Resources
We list only the title, our own summary, a link to the original and the author’s name. We never republish the content. Know something good? Use the submission form and choose “Recommend a link”. If you are the author and want an entry changed or removed, use “Request a correction or takedown” at the bottom of the page.
Explore Chinese and English tutorials, app integrations, skills, routines, plugins and bot templates. Filter by source language or search for an app such as Notion or Gmail.
Explore the 24-source reading map for the rewritten workflows, role cards and actual verification scope.
Practice with the complete tutorial library
Browse all 25 complete tutorials →Start with original practice inputs, follow the staged prompts, then compare the complete reference results and acceptance checks. Each tutorial states what was tested and what still needs validation.
391 resources
No matching resources. Try another category, platform, source language or search term.
Chief-of-Staff-led bot org: approvals locked first, Notion as shared brain (Start Here page, context database), morning brief, three-layer org chart, hiring prompt, top-10 rules, six-part SOPs, 30-day rollout.
AppsGrok BotNotionWhat was checked
Read the article structure and main first-half content (~38k chars; later safety and rollout sections at heading level only); images not viewed. Author bio: former AppSumo CEO; figures self-reported. Distinct from a similarly titled maestrooth guide. Not reproduced.
Seven ways to stretch the weekly pool: which routines to pause, cadence, narrower triggers, group-chat size, bot-to-bot messaging, starting small with before/after examples, and meter checks; plus when not to cut.
AppsGrok BotWhat was checked
Full X Article read; images not viewed. Cites xAI docs; plan prices and multipliers as of Nov 2 may have changed. The author's Nov 9 promo quote is merged. Not reproduced.
Step-by-step one-person-company guide: one job per first bot, minimal logins, teach once, schedule, a reviewer bot; five starter bots; and a QA maturity arc from 100% review to 50% and 10% spot checks.
AppsGrok BotWhat was checked
Full X Article read incl. charter and prompts; images not viewed. Author is an agency operations advisor; revenue and margins self-reported; bio links to his services. Not reproduced.
Twelve cold-email prompts: before sending (offer test, list, inbox math, angles, spam audit) and after (inbox health, fast reply drafts, sorting, rotation, 10k-email call, rules). The bot never sends.
AppsGrok BotInframailWhat was checked
Full X Article read incl. 12 prompts; images not viewed. Author is Inframail's co-founder/CEO; revenue and volume claims are self-reported and involve his product. Follow local email laws. Not reproduced.
After a 16-bot roster burned the weekly limit in 4.5 days, the author analyzed the usage CSV, found cached context reloads and four routines drove most burn, paused seven routines, and added a morning meter check.
AppsGrok BotClaudeWhat was checked
Full post text read. Figures are the author's own (analyzed with Claude); CSV not seen; plan choices as of Nov 7. Not reproduced.
Twelve usage-saving settings: one chat per task, wider schedules, narrow triggers, pausing, a routines bot, small tests, fewer tools, ask-first rules, offloading coding, a cap, meter checks, tight briefs.
AppsGrok BotWhat was checked
Full X Article read incl. copyable rules and brief template; images not viewed. Usage figures are the author's own; promotes his Substack. Billing details as of Nov 4. Not reproduced.
Tested setup: webhook-triggered routine plus an iPhone Shortcuts message automation (sender + keyword) that POSTs the message text as JSON. Lists example GROK commands, which the author says were not tested.
AppsGrok BotiPhone ShortcutsiMessageWhat was checked
Full post text read; images not viewed. Author tested only the iMessage trigger; example commands untested. Keep the auth header private and restrict the sender. Not reproduced.
Webhook-triggered daily briefing routine fired by an iPhone Shortcuts 'alarm stopped' automation POSTing JSON with an Authorization header; suggests other triggers like CarPlay, arriving home or NFC tags.
AppsGrok BotiPhone ShortcutsiPhone ClockWhat was checked
Full post text read; images not viewed. Author reports testing the alarm trigger; other triggers are ideas. The author's Nov 8 quote post is merged here. Not reproduced.
Student-founder life team: startup, study, health, life and reflection bots under a Chief of Staff that alone writes the daily list; 8am plan, evening and Sunday reviews; sends, spending and posts need a yes.
AppsGrok BotWhat was checked
Full post text read; the closing link and images not opened. New, small account; self-reported personal case. Not reproduced.
Real estate owner's 15-bot setup: one job per bot, a Chief of Staff hub, Granola webhook handoffs of call transcripts, skills ported from Claude, approvals for sends, ad budgets and posts, a backup harness.
AppsGrok BotHighLevelSkoolMeta AdsYouTubeGranolaClaudeHermesWhat was checked
Full post text read; closing link and images not opened. Self-reported case; third-party tools not reviewed. Not reproduced.
Assigns Host, Progress, Improve and Execute roles through one owner, five-part briefs, evidence lanes, project profiles and quota failover. Its MCP/skill layer is portable but does not replace host approvals or tools.
AppsGrok BotCodexWorkBuddyWhat was checked
README, framework and version notes read; MCP not installed or run in any host. It is explicitly only an orchestration layer. Historical example paths, host tools, permissions and quota failover require local adaptation.
Five skills cover gene specs, evidence transcription, proof gates, legacy proposals and research. Each round starts with a task contract and independent checks; humans decide goals, external writes and acceptance.
AppsGrok BotWorkBuddyCodeBuddyWhat was checked
Chinese README, skill list and validation notes read; not installed or run. Local validation is described, but real multi-agent WorkBuddy/Grok Bot operation was not tested end to end; outcomes and safety are not implied.
Rename existing Bots by job, pair them with 35 SVG/PNG icons, confirm changes in each Bot chat, and verify the sidebar. Optional guides audit overlapping roles and explain that consolidation does not migrate history.
AppsGrok BotWhat was checked
README, catalog, prompts and confirmation rules read; no Bot was renamed or re-identified. This is a documentation/asset kit. Supported actions may change, and only user confirmation plus visible sidebar results count.
Scout, Curator, Bargain and a main Bot discover, score and price events in stages, send zero to three nightly picks, and update a five-tab Page. Calendar access is read-only, with no booking or payment.
AppsGrok BotCalendarPagesWhat was checked
README, structure, data and permission boundaries read; template not imported, routines not run and calendar not connected. Locks, backups, schemas, atomic writes and privacy scan are author-provided; example prices need rechecking.
Select an existing Bot in Raycast, review a draft, then explicitly send its webhook. The local tool keeps request IDs and delivery states, not message text, avoids blind retries, and leaves replies in Grok Bot.
AppsGrok BotRaycastWhat was checked
README, setup boundary and receipt semantics read; not installed and no webhook sent. It requires macOS, Raycast, Node 24.11+ and a Bot webhook Routine. Local credential handling and tests are author-reported.
A CLI/MCP checks requirements, scope, current evidence, unknowns and calculations, with optional independent review. Prior approvals are never reused, and pass never authorizes real-world actions.
AppsGrok BotCursorxAI APIWhat was checked
README, states, configuration and limits read; MCP not installed or registered and no external review called. Offline tests and smoke checks are author-reported. It was not tested inside Grok Bot and only gates packets sent to it.
An OAuth MCP searches call recordings and transcripts stored in the user's Google Drive, creates after-call notes, triages an inbox and builds project digests. Delete, settings and webhooks require extra permission.
AppsGrok BotCallboxGoogle DriveGoogle MeetZoomMicrosoft TeamsWhat was checked
README, package structure, permissions and limits read; no Callbox or Google account connected. The vendor says assistants cannot start recordings, permanently delete, change Google connections or delete accounts; server and data handling were not audited.
Connects a study skill and OAuth MCP to Grok Bot for cited syllabus answers, one-question quizzes, progress and graded mock exams. It needs a Statsprøven account and is not an official exam service.
AppsGrok BotStatsprøvenWhat was checked
README, connection steps and product limits read; not signed in and no mock exam run. Quotas, server grading and answer handling are vendor claims. Cursor desktop is unsupported; official exam information remains with HK-dir.
Claims a Bot inbox and schedules an 8 p.m. brief, then distills chats and files into skills synced to private GitHub. Pages hold reports and boards; finder and video-director Bots handle retrieval and clips.
AppsGrok BotGitHubPagesWhat was checked
The original YouTube ID confirmed title, author, publish date and 486-second length, cross-checked with the creator's public description and index. No usable captions were available; the workflow was not reproduced.
A public log of one author and Grok Bot building Frame spaces, shaping environments and wiring integrations, with methods, limits, roles, an undo plan, OpenAPI spec and a read-only example.
AppsGrok BotFrame VRWhat was checked
README and public index read; Frame was not connected and no API called. The repo labels documented/tested claims Confirmed and untested ideas Proposed; execution results remain author-reported.
Keeps a Grok Bot cloud computer active until a bounded deadline, watches existing or requested processes, reports completion, and shares one watcher. The author reports roughly 2–3% usage for one five-hour test.
AppsGrok BotWhat was checked
The trilingual README, capabilities and disclaimer were read; not installed or run. Keep-alive behavior, the 24-hour ceiling and usage figure are author-reported and remain platform-dependent.
Separates builder, primary reviewer, independent QA and final decider, using JSON-Schema handoffs, frozen rubrics, a SHA-256 chained event log and tool/round/time limits to prevent self-approval and log escalations.
AppsGrok BotNotionGitHubWhat was checked
README, layout and examples read; scripts and Notion were not used. A hash chain can reveal log changes but cannot prove event truth; named roles and the Notion page require adaptation.
Uses Imans OAuth/MCP for authorized sales, stock, finance, marketing and process data, plus tasks, process drafts and approved price publication; tools vary with workspace roles, connection mode and server discovery.
AppsGrok BotImansWhat was checked
README, permissions and release status read; no account or tool used. Direct MCP is vendor-reported working, while packaged listing acceptance remains pending. Future tools are discovered and some operations consume credits.
Polls a same-origin mailbox snapshot every 30 seconds for routing and receipt states. Invalid, failed or stale snapshots clear; public export keeps reviewed metadata without bodies or credentials.
AppsGrok BotMeta MuseWhat was checked
README and boundary notes read; no private mailbox or producer used. The dashboard cannot refresh the private source, execute tasks, send ACKs or wake agents; the ten-minute producer schedule is separate setup.
Adds a chief and replaceable specialists to Hermes, delegates long work to the user's Grok Bot via HTTPS webhook or shared inbox, and collects results. It is neither migration nor a Grok Bot API.
AppsGrok BotHermesWhat was checked
README, tools and security notes read; not installed and no webhook sent. HTTPS, redirect/userinfo rejection, file limits and no background work are author claims; source and end-to-end Routine were not fully audited.
A seven-Skill beta compares jobs with resume evidence and gaps, drafts factual resumes and cover letters, and offers optional heuristic scoring. Each application action needs exact approval.
AppsGrok BotStitch Your Pitch MCPWhat was checked
README, version and availability notes read; template not added and no resume, sign-in or application used. The backend is closed, new-user isolation checks remain open, and scores are not ATS or hiring predictions.
Turns text, voice, sketches or photos into a live prototype, SPEC, private GitHub repo, research/docs and an MVP PR, with Grok Bot driving local MCP tools for builds, shipping, reminders and memory.
AppsGrok BotCursorGitHubxAI APIWhat was checked
README and GROKBOT.md read; not run. It needs an xAI key and GitHub repo access; Cursor builds may request all-repository app access. The local service, source and data handling were not fully audited.
A local Windows service watches folders, clocks, clipboard or on-device wake phrases, opens a named Bot through grokbot:// and puts a brief on the clipboard; no Grok Bot usage occurs until the user sends it.
AppsGrok BotWindowsWhat was checked
README, requirements and test notes read; not installed on Windows. It does not auto-send. Local file watching, microphone permissions and protocol-link compatibility were not independently reproduced.
A macOS companion saves Grok-authored HTML and state, with five CLI families, atomic publishing and revision checks. Pages are network-free and sandboxed; local handlers run allowed actions.
AppsGrok BotmacOSWhat was checked
README, native route and security boundary read; not built or run. A private native Page was author-tested on 0.79.2, but no persistent Page-to-Bot bridge exists. Handlers run as the current user without an OS sandbox.
Prove a read-only check first, then confirm owner, timezone and When to run. Validate with Test or the same manual task; troubleshoot Paused, schedule, timezone, first run and failures instead of recreating Routines.
AppsGrok BotWhat was checked
The public article and author metadata were read. Its official-document references were not replayed step by step; no Routine or account connection was created and no usage was spent.
Connects SimpleVox business numbers to Grok Bot for missed-call follow-up, transcript search, drafted/approved texts, assisted calls, contacts and reports, with explicit confirmation and idempotent retries.
AppsGrok BotSimpleVoxWhat was checked
README, permissions, safety notes and license were read. Nothing was installed or connected; no customer data, text or call was touched. Service, privacy and approval behavior remain vendor-reported.
Two-part checklist: fresh task chats, batched updates, slower idle Routines, narrow triggers, short handoff files, a dedicated Routine Bot, connectors before browsing, scripts after validation, and audit prompts.
AppsGrok BotSlackGmailGoogle CalendarCloudflare WorkersWhat was checked
Both Reddit posts were read. Images, cited videos and external sites were not checked; claimed 35–75% savings were not verified, and no installs, migrations or automations were run.
Import Dr. Eggbot, let it interview you, give each bot one job/voice/tools plus anti-jobs, put a lead that only routes, weekly transcript audits, and a fix log. Reply has a pasteable starter prompt.
AppsGrok BotWhat was checked
Main note_tweet and author reply read; images not viewed. Quoted older article not fully re-read. Not reproduced.
Explains Memory vs Skill timing: put Job, Anti-jobs and fixed URLs in Memory; keep SOPs in Skills. Includes a pasteable anti-drift Memory skeleton and a clean-before-share warning.
AppsGrok BotWhat was checked
Full X Article text read; screenshots not viewed. dr eggbot comparison is author-reported; not reproduced.
Airbnb bot passed a cleaner’s toilet-paper shortage to a house-manager bot, which edited an existing HEB delivery order—no owner involvement.
AppsGrok BotWhat was checked
Full text read; chat logs and connector setup not inspected. Single author case; not reproduced.
Daily job agent: review profile, find openings, filter by rules, apply matches, log to Google Sheets (~45 in 7 days). Author flags rising token use.
AppsGrok BotGoogle SheetsWhat was checked
Full text read; sheet and apply-site automation not verified. Author-reported; not reproduced.
Paste roster for start/sit, pull injury news before kickoff, set a Routine to ping if anyone is ruled out—TNF third-string QB context.
AppsGrok BotWhat was checked
Full text read; injury source is a quoted post, not independently verified. Routine not reproduced.
Plan choice, install/sign-in, and first-message “remember” for job+rules. Distinguishes which Settings panes affect all Bots.
AppsGrok BotWhat was checked
Original language is Japanese; the catalog currently supports zh/en only, so this is filed under en and tagged source-language-ja. Full X Article text read; author verified on 2026-11-04 Windows desktop 0.78.x; images not viewed; not reproduced.
Five-part task asks (outcome, sources, bans, deliverable, checkpoint). Explains approval-card buttons and always-ask rules in Permissions.
AppsGrok BotWhat was checked
Original language is Japanese; the catalog currently supports zh/en only, so this is filed under en and tagged source-language-ja. Full X Article text read; author verified on 2026-11-04 Windows desktop 0.78.x; images not viewed; not reproduced.
Open the shared cloud PC, take over/stop, vault for secrets, and safe recovery order. Stresses one PC shared by all Bots.
AppsGrok BotWhat was checked
Original language is Japanese; the catalog currently supports zh/en only, so this is filed under en and tagged source-language-ja. Full X Article text read; author verified on 2026-11-04 Windows desktop 0.78.x; images not viewed; not reproduced.
Installed vs connected, scope equals account rights. Three X paths (public reads, X plugin, @bot) and disconnect cleanup order.
AppsGrok BotWhat was checked
Original language is Japanese; the catalog currently supports zh/en only, so this is filed under en and tagged source-language-ja. Full X Article text read; author verified on 2026-11-04 Windows desktop 0.78.x; images not viewed; not reproduced.
Teach a task: record once, answer the goal prompt, save skill. Review the 6-point draft, add failure handling; no secrets while recording.
AppsGrok BotWhat was checked
Original language is Japanese; the catalog currently supports zh/en only, so this is filed under en and tagged source-language-ja. Full X Article text read; author verified on 2026-11-04 Windows desktop 0.78.x; images not viewed; not reproduced.
Import dr eggbot, state you want to share, one-job bot, strip PII and host UI on GitHub, say “make a template” then Publish, re-pack on updates. Lists practical publish tips.
AppsGrok BotWhat was checked
Full article text read; screenshots and live template pages not opened. Author-reported X Pulse flow; publish not reproduced.
Invite-only; no public form. Eligibility: 18+, X Money U.S. state, good standing, verified email, eligible Premium, X Money for payouts. Links help center + one creator story.
AppsGrok BotWhat was checked
Full post read; help-center page not re-fetched in full. Eligibility paraphrases official copy; no application attempted.
Templates are recipes not clones: Add to Grok Bot, preview, then join. Three steps: one job per bot, bounded first task with review, automate only after trust.
AppsGrok BotWhat was checked
Post text read; attached video not watched. Add-template flow not reproduced.
Five community cases: multi-shop Xianyu ops, template rewards pilot, inbound ad sales triage, pre-sales replies, Meta/SEO assist—with caveats on invite-only rewards and estimated ad ROI.
AppsGrok BotWhat was checked
Main post text read; images and cited Reddit/source posts not verified for revenue claims. Author-reported roundup; not reproduced.
Three tips: egress via your device network, prefer browser over brittle scripts with human publish, and test Skills/MCP on a small account first.
AppsGrok BotWhat was checked
Full text read; platform risk outcomes not verified. Author tips; not reproduced.
Compares cloud schedules, desktop tasks and in-session loops across four tools, including clock vs event triggers, machine availability, permissions, approval boundaries and a repository-monitoring example.
AppsClaude CodeCodexCursorGitHubSlackLinearSentryPagerDutyWhat was checked
Full article and citations read; no task created. Products change, so triggers, intervals, plans and UI should be rechecked in official docs; test buttons may perform real actions.
Create a Grok Bot webhook Routine, then let a local Claude Code Skill send requests and poll reply files with a standard-library script; the example reads the Following feed and filters it into a linked digest.
AppsClaude CodeXWhat was checked
Full article read; not reproduced. The author reports a 7–8s pong, but the Skill/script is private. 194 posts covered about 2.5 hours, costs were estimates rather than bills, and X capabilities/pricing need rechecking.
Add Reason's custom MCP to Grok Bot, sign in, and join a local Claude Code session to the same network. The Bot can list sessions, dispatch work, receive replies, and use a Routine to wake on a response.
AppsClaude CodeReason NetworkWhat was checked
Full guide read; not connected. It says updated Nov 2026 without a day. Reason is a closed commercial service; local machines must stay online, cloud sessions need a plan, and Bot handoff messages use the weekly pool.
An eight-Bot ops template enforces one job per Bot and stops only for send, money, delete, credential, brand or fact decisions, with a reviewer, validated run logs, routing, low-cost prechecks and retirement rules.
AppsGitHubNotionLinearTrelloWhat was checked
README, tree and license read; not installed. The template needs a public repo or read connector, placeholders and privacy review. Safety and savings are design goals, not independently tested outcomes.
A loopback helper exposes MCP through a bearer-protected Cloudflare tunnel. It lists sessions by default; opt-in flags enable starting, following up, reading or stopping Codex tasks, with optional root limits.
AppsCodexCloudflare TunnelmacOSWhat was checked
Safety and uninstall docs read; not installed or run. Anyone with URL and token can read sessions and, with flags, run tasks. Tunnel URLs change, logs may be sensitive, and the code was not fully audited.
The author had Grok Bot direct Claude Code on its cloud computer to add page history, note filtering and reading progress to Folio, recording typecheck, build, logic-test and smoke-test results.
AppsClaude CodeGitHubWhat was checked
Repository and notes were read; not run. Tests and collaboration are author-reported, the UI was not visually verified, and OAuth setup or live sessions were not independently checked.
A dependency-free Node CLI turns local-agent work into JSON tickets, wakes a Grok Bot Routine by webhook, and manages duplicates and failures with idempotency, receipts, reply files, rate caps and a kill switch.
AppsClaude CodeWhat was checked
README, Skill, examples and license were read; nothing was installed, run or posted. Compatibility, security and end-to-end receipts remain author-reported.
Routes Slack DMs or threads through Socket Mode or a local/Cloudflare relay into a Grok Bot webhook Routine, then returns answers, attachments and voice output to the source thread with sharing controls.
AppsSlackCloudflare WorkersntfyWhat was checked
README and tree were read; Slack was not connected or deployed. This is a one-commit public release with no declared license; reliability, permissions and attachment handling were not independently tested.
A macOS tool records a held-key voice question and screenshot, sends the prompt plus a temporary image link to a Grok Bot webhook, and returns the answer through ntfy for screen Q&A without changing windows.
AppsmacOSimgbbntfyWhat was checked
Source docs were read; not installed or run. The app is not notarized or tested on a second Mac. Screenshots use a public imgbb link for about ten minutes, and the ntfy topic must be treated as a secret.
A CV Skill uses a Master Resume as the sole fact source, checks employment completeness before output, fails closed on missing facts, and includes a short setup and contribution template.
AppsClaude CodeCodexWhat was checked
Skill, templates and license were read; not installed or tested on a real CV. It constrains the workflow but cannot establish that facts are true or replace human review.
An Electron desktop app unifies quota and reset views: Grok Bot/Cursor through a Cursor session, ChatGPT through Codex app-server, Claude from local records, and OpenAI through an optional Admin key.
AppsCursorChatGPTClaudeOpenAI APIWhat was checked
Security docs and representative source were reviewed; not installed or fully audited. It opens Cursor's database read-only, calls the OS secret store and can save an OpenAI Admin key. The app is unsigned.
Eight Skills cover trust ladders, product verification, Skill evals, CI or architecture constraints, five-part briefs, isolated autonomous runs, team loops and manual-fix-rate measurement.
AppsClaude CodeCodexMuseWhat was checked
Repository, sources and license read; not installed. It is an independent synthesis of Lauren Tan's public work, not official or endorsed; cited output figures are unaudited and methods were not independently reproduced.
A self-hosted dashboard turns an append-only JSON team log into briefings, approvals, work-in-progress, deadlines, usage and Bot cards, with passkey sign-in, Oracle Cloud VM deployment and a setup Skill.
AppsOracle CloudGitHubWhat was checked
README, schema, deployment and test notes were read; not run. Security, free-tier availability, uptime and team outcomes were not independently verified, and deployment creates external cloud resources.
A remote-MCP plugin gives read-only access to Priosweep, with three Skills to prioritize dependency updates, explain a finding and produce a current security briefing across Grok Bot, Cursor, Claude Code and Codex.
AppsCursorClaude CodeCodexChatGPTWhat was checked
Repository and MIT license read; service not connected. It is not yet listed publicly. Priosweep privacy and terms pages were inaccessible this round, so remote-service scope, OAuth data and accuracy need separate review.
A Python engine evaluates Agent Skills, plugins and MCP servers across 12 dimensions and emits JSON/Markdown/HTML reports; 19 Grok Bot Skills orchestrate judging, routing, security, sandbox pentests and GitHub drafts.
AppsClaude CodeCodexGitHubWhat was checked
README and limitations read; not run. The 42/42 self-test uses co-designed seeded flaws and is not real-world recall. Docker CI, remote MCP/OAuth and cross-model consistency remain incomplete.
Four rules require completed deliveries to end with “Work is completed,” add a “Work in progress” note when anything is still queued or running, and reserve progress-only messages for the in-progress marker.
AppsCursorWhat was checked
The post and author replies were read. The rules are visible, but effectiveness is anecdotal; no project or error-rate comparison was run, and comments still stress human quality control.
The author says long chats caused about three quarters of one week's usage, recommending batched asks, short answers, narrower browser/research scopes and less frequent Routines; Bot handoffs charge both sides.
What was checked
Post and comments were read; billing and the shared Bot were not checked. Numbers are account-specific. A comment notes there is no normal new-chat button, so separate Bots or blocked-message Reset are more accurate.
A 20-chapter Chinese guide with 21 prompts: build a traceable brief, offload approved material through Pi to MiniMax or DeepSeek, then save a Skill, schedule a Routine, hand off between Bots and build an ecommerce brief.
AppsPiMiniMaxDeepSeekWhat was checked
The page and version notes were read; Pi, keys and end-to-end flow were not tested. The author states the integration was not tested in readers' accounts. Models, prices, permissions and UI can change.
A self-hosted MCP keeps preferences as Markdown in the user's private GitHub repo. Agents read the same rules and only propose edits; passkey approval creates one evidenced commit, with per-connection category scopes.
AppsGitHubCloudflare WorkersClaude CodeCodexChatGPTWhat was checked
README, tree and security entry points were read; not deployed or connected. Grok Bot support and testing status are author-reported. Setup needs Cloudflare, a fine-grained GitHub token, OAuth and a private repo.
An MIT Skill estimates green, yellow or red from chat length and compaction, keeps STATUS.md and a handoff file, and suggests a fresh Bot on red; it also includes a local dashboard and optional BotDuplicator flow.
AppsBotDuplicatorWhat was checked
README and English Skill were read; not installed or run. Thresholds are rough estimates, not token counts. The optional Live Status connector is not deployed, and handoff files need privacy review.
Aqua Voice's MIT client config connects Grok Bot to its remote MCP to search saved dictations, read transcripts, manage dictionary, replacements, instructions and language, and submit requested feedback.
AppsAqua VoiceCursorWhat was checked
README, manifests and license were read; no account connected. The repo contains client config, not hosted backend source. History depends on privacy settings, and OAuth may request read and write scopes together.
A sanitized blueprint with 43 Bot profiles, four managers, 55 Routines, connectors and shared files. Work is read- or draft-only by default, writes need owner approval, and a manager chain reduces duplication.
AppsSalesforceSlackGmailGoogle DriveNotionGitHubCloudflareWhat was checked
README, guardrails and Routine table were read; no systems connected. The author says schedules came from profiles and shared memory, not automation records. No license is declared; live results were not verified.
An MIT tool imports role, Skills, memory and Routine descriptions into Hermes, with two-way sync, conflict detection and restore points. Optional g2h messaging frames peer text as untrusted data.
AppsHermes AgentWhat was checked
README, tree and security entry points were read; not run. This is an early v0.1 with one maintainer; live import and 10/10 exchanges are author tests. The bridge is not a security boundary and bundles contain personal data.
An open bilingual daily column with four issues on Bot management, Routines, Skills, cloud computers and connected apps, each with copyable prompts, a worked example, a 15-minute exercise and source links.
What was checked
Read the README, MIT license, repository tree and Nov 4 Chinese source. Site and build were not run. It is an unofficial editorial project largely based on official docs; recheck changing UI and limits.
Fourteen bilingual personal-assistant roles and group-chat patterns cover secretary, finance, legal, health, tutoring, translation and content, with prompts, placeholders, output rules and some cron Routines.
AppsGmailWhatsAppFacebookXWhat was checked
README, manager and health templates were read; dependencies were not installed or tested. No license is declared. Legal, health, tax and investment roles cannot replace professionals, and sensitive data should stay out of shared memory.
Five MIT recruiting packs contain 50 SKILL.md files for agencies, in-house teams, outreach and growth. A shared profile grounds facts; the Skills draft and check, but do not screen, rank or make hiring decisions.
AppsLinkedInEmailWhat was checked
README, tree and two sample Skills were read; not uploaded or tested. The repo targets Claude first and says SKILL.md can be pasted into Grok Bot; compatibility and local hiring compliance need verification.
Two-way handoff between desktop WorkBuddy and Grok Bot via a shared Feishu doc and a webhook Routine; results are appended by number, never overwritten, and read back by number. WeChat-log case and three prompts.
AppsWorkBuddyFeishuWhat was checked
About 3,100 characters of article text read; the command section at the end was missing from plain text and not obtained this round; images not viewed. Involves third-party CLI auth and webhook keys.
A 30+ bot company: talk only to a chief-of-staff bot, bots report up and work in group chats, connected to mail, calendar and CRM, run daily and weekly Routines, and follow six guardrails like approve-before-send.
AppsHubSpotTwilioSlackGmailWhat was checked
Full article plain text read (about 6,200 characters); images not viewed. Phone-calling and trading results are self-reported; ends with a consulting promo.
Build one style-guide Skill per platform: list about seven writing traits plus your own sample posts, let the Bot draft, then write every correction back into the Skill so output converges on your voice.
What was checked
Article text read; images not viewed. The example-prompt section is empty in plain text (likely images), so the exact prompts were not obtained this round.
Part 2 of a long guide: 10-step setup (phone apps, Vault, Action Button, Primary Bot), life-admin Routines (morning Page, draft-only inbox triage, subscription cleanup), about 30 hidden features and a prompt pack.
What was checked
Full article plain text read (about 17.5k characters); images not viewed. Upcoming-feature claims (1.0, texting) are the author's reports of rumors and unverified; part 1 was not obtained this round.
A Skill that compresses docs to 10–15%: extract, don't paraphrase (decision, rejected options, checkable claims, open items, numbers, owners), keep the author's items apart from your concerns, keep identifiers verbatim.
What was checked
Read the author's 7-post thread; the Skill file itself was not posted and was not obtained this round.
Refresh a long-running bot: the old one writes a handover file, a fresh copy saves only lasting rules and recreates Routines verbatim, replies DONE, old Routines are removed, a human deletes the old bot. Two prompts.
What was checked
Full article plain text read; images not viewed. The 30–50% saving is the author's estimate without a stated measurement.
Features-to-onboarding tour: send, pay, delete or trade always come back to you; every Routine wake costs, so prefer events, slow schedules and silence when unchanged; four levels and a 7-step first week.
AppsCursorWhat was checked
Article text (about 2,800 characters) read; the 38 slides in the thread are images and were not viewed. Ends with a meetup promo. Same-author thread replies merged here.
The author's five-step loop for running Chinese and English X accounts with Grok Bot: find the problem users urgently need solved, write as hook, proof, steps, catch and close, then review saves and reposts to adjust.
AppsXWhat was checked
Full long post read; images not viewed; results are self-reported.
Agent Recorder captures Grok Bot's own Linux X11 display before browser work, then returns an MP4 and action summary. FastFileLink can delete after download, or upload can use Google Drive; the Skill pins v0.2.2.
AppsGitHubFastFileLinkGoogle DriveFFmpegWhat was checked
Article, README and pinned Skill were read; installation, capture and transfer were not tested. It does not record the user's Mac/Windows and is not tamper-proof. Recordings may expose personal, admin or key material.
A CC0 English library splits getting started, learning path, capabilities, limits, Bots, Skills, Routines, connectors, Pages, shared computer, starter roles, prompts, safety and troubleshooting into source-linked docs.
AppsGitHubWhat was checked
README, tree and document index were read, with spot checks of major official links; 20-plus docs were not audited line by line. It overlaps Chinese guides but differs in language, CC0 licensing and modular checklists.
An MIT text kit with org structure, seat cards, approval gates for sends, spending, deletion and deploys, traceable handoffs, low-turn coordination, plus change-only watcher, keep-alive and webhook-dedupe Routines.
AppsGitHubWhat was checked
Core rules, handoff and credit-saving docs plus the template tree were reviewed. No Bots or Routines were created. Strong rules need local adaptation, and template approval inheritance is not a platform permission mechanism.
The Skill loads self and comp listings, reads an AirROI key from 1Password, saves snapshots, compares weighted rates and occupancy, then proposes pricing actions within history, legal-stay and call-budget rules.
AppsAirROI1PasswordGitHubAirbnbVRBOWhat was checked
README, tree and Skill were read. AirROI was not purchased or called, secrets and real listings were not accessed, and results were not validated. The author lists about $0.20 per listing pull; real price changes need human review.
A workshop case uses four Grok Bot Routines for briefing, rating watch, stress pulse and on-demand sentiment across 12 Thai listed firms, publishing JSON/Python outputs to GitHub Pages with a public-data-only disclaimer.
AppsGitHub PagesPythonWhat was checked
README and repository structure were reviewed. Routines/scripts and company data were not tested, and the Nov 4 workshop was not attended. This is a training workflow, not investment, credit or client advice.
Three Skills create two art options from a photo, trace the chosen one into a CNC-ready SVG, render a cut preview, and return dimensions, path count, rough timing and CAM notes. Examples and MIT license are included.
AppsGitHubPythonCAMWhat was checked
README, tree and the end-to-end Skill were read. Python code, dependencies, image generation, scripts and machining were not tested. Timing and tooling are author estimates; CAM simulation and machine safety take priority.
A Chinese guide covers setup, connectors, Routines, Skills, multi-Bot work, cloud computer, templates, cases, quota, safety and pitfalls, labeling sections as verified or author-claimed and linking primary sources.
AppsGitHubWhat was checked
The full README and its claim labels/source structure were read; community cases were not reproduced. GitHub reports NOASSERTION while the README declares CC BY-NC 4.0. It overlaps Chinese guides but uses a distinct evidence-labeling scheme.
The dashboard shows each Bot's work, status and link plus a ranked “Waiting on you” list. Focus for Today ranks tasks from email, calendar, Bot updates and chat. Both update one Page and stay quiet on no change.
AppsPagesGoogle CalendarEmailWhat was checked
The full thread and both comment templates were read. No Page, inbox/calendar connection or Routine was created. The templates prohibit guessing and secrets, but real connections still need least privilege and human review.
A travel eSIM project uses Boss, Supplier, Shopify, Design, Marketing, Service and Accounting Bots. They retain role context and hand off, while a human reviews every send. It is week one, with no launch or sales.
AppsShopifyEmailWhat was checked
The public post was read; only image alt text was available. Store, suppliers and handoffs were not verified. The author says Bots can be confidently wrong, every send is human-reviewed, and there are no business results yet.
A Writer Bot lays out a public Canvas edition; Dispatch emails its link only after “send it”; a Classic edition stays private. The pattern separates layout from distribution and personal data from the public copy.
AppsPagesEmailWhat was checked
The public post was read; the sample, source quality and email delivery were not tested. This is a short case rather than a step-by-step tutorial, and its privacy split does not replace a human pre-publication review.
Five Bots cover research, writing, review, inbox and daily digest. Run each job once, save it as a Skill, then add a Routine; the guide also covers group-chat kickoff, quota, draft approval and shared-computer limits.
What was checked
Tie Zhu read the full article and obtained six code blocks; Codex checked the handoff and deduplication without independently reading X. Images and each product-limit claim were not checked; not reproduced.
Use one Page per project plus a master index. Bots read before work and update afterward, tasks use short chats, each project gets a weekday heartbeat, notifications are exception-only, and closed projects are archived.
AppsPagesWhat was checked
Tie Zhu read the full post and prompt; Codex checked the handoff and deduplication without independently reading X. Not reproduced; heartbeat Routines consume quota and the author suggests a seven-day single-project trial.
- Grok Bot allowance gone in two days? Reduce waste with a private entry point and mobile HTML reports
Add a private entry point to the manager Bot so a group message does not wake every Bot, and replace long reports with mobile-friendly HTML, optional screen recordings and a final list of decisions needed from the user.
What was checked
Reviewed the public Bilibili description only; the video was not watched and usage savings were not reproduced. Allowance figures are account- and plan-specific author reports, not general results.
Split affiliate work into research, program checks, drafting, link tracking and price monitoring; require sourced facts while a human verifies prices, adds disclosures and approves publishing.
AppsFirecrawlWhat was checked
Read the full article and prompts; not run. The author checked official docs Oct 28 and rejects income guarantees, but repeatedly promotes AIM Agent, courses and affiliate links, creating a direct commercial interest.
Use an idea Bot to group sourced questions, build a voice Skill from writing samples, split research, drafting and repurposing across roles, and keep publishing behind human approval.
AppsYouTubeRedditXHiggsfieldWhat was checked
Read the full article and prompt; not run. Pages, media limits and mobile capabilities are the author's Oct 28 reading of official docs, not independently tested. The article promotes AIM Agent, courses and affiliate links.
Seven Bots read versioned rulebooks while GitHub Actions advances a fixed state machine from Slack intake through Cursor agents, independent acceptance and review, with human confirmation at key gates.
AppsSlackGitHubGitHub ActionsCursor Cloud AgentsWhat was checked
Read the README, setup, rulebooks and state machine; not run. All data is fictional. Deployment needs Slack/GitHub permissions, webhook keys and a classic public_repo token; permissions and router code were not security-audited.
Four Skills collect rig settings, plan paced routes and stays, build a map board, and independently check route, fit, fuel, stays, weather, budget and bookings while stopping before sending, booking or paying.
AppsGitHubOpenStreetMapTwilioWhat was checked
Read the README, four Skills, version and changelog. Nothing was imported or run. The monthly updater fetches Skills from main, so upstream changes are a supply-chain risk; the Twilio caller is optional and advanced.
A community thread covers family hubs, job search, workouts, bills, school notices, smart homes and sports research, including an authenticated Apple Health webhook and workout logging to a sheet.
AppsApple HealthGoogle DriveGoogle CalendarGmailYNABActual BudgetSleeperHome AssistantEcobeeGarminTodoistWhat was checked
Read the post and representative detailed comments, not every reply; no outcomes were reproduced. Health, financial, child and family data are sensitive. Inclusion does not recommend granting the same access, and security claims were not verified.
A searchable Chinese handbook covering first tasks, cloud computer, Routines, Skills, usage, real cases, multi-Bot teams, webhooks, iOS shortcuts, CLI/MCP, safety tiers and troubleshooting, with prompts and checks.
AppsGmailSlackFeishuWeChatGitHubWhat was checked
Read the structure, representative content, sources and changelog; not every external link was reopened and workflows were not reproduced. Community claims are labeled unverified; official material was rechecked Nov 2. No explicit repository license.
Choose a missing role and a leader with many public videos, distill transcripts into a sourced playbook, add a non-endorsement caveat and ownership limits, connect real data first, and keep one or two weekly checks.
AppsYouTubeWhat was checked
Tie Zhu read the full post and article; Codex reviewed the handoff without independently reading X. Images not viewed; not reproduced. Contains an unedited placeholder and promotion; imitation is not endorsement.
Daily prompt: check @poteto's new posts, compare them with current Bots, Skills and Routines, report only useful changes with links and paste-ready edits, change nothing until confirmed, and stay quiet otherwise.
AppsXWhat was checked
Tie Zhu read the full prompt; Codex reviewed the handoff and deduplication without independently reading X. The quoted poteto post is reviewed separately; not reproduced.
Covers chief-of-staff delegation, clear roles with drafts for outbound actions, a prompt to migrate local CLAUDE.md and Skills, first-day connector groups, a 7am research Routine and a 7-day plan.
AppsClaude CodeWhat was checked
Tie Zhu read the full post and article; Codex reviewed the status change without independently reading X. Not reproduced. Dr. Egg Bot's role conflicts with poteto's description, migration asks for always-allow local execution, and the article promotes a community.
Record a weekly browser chore once, let the Bot draft a Skill, then replay it on last week's data, list every step where it guessed and ask about each one to fill missing decision rules.
What was checked
Tie Zhu read the post; Codex reviewed the handoff without independently reading X. Attached media not viewed, the recording limit is author-reported, and the workflow was not reproduced.
Before rebuilding a Routine, check that it is enabled, its schedule and time zone, whether its trigger is too narrow or noisy, and the Test result; or ask the Bot to diagnose these and change nothing.
What was checked
Tie Zhu read the post; Codex reviewed the handoff without independently reading X; not reproduced.
Extends poteto's guide with a seven-question role card, permission tiers, a scored friction log, required Routine fields, a research-editor-growth team case, a 7-day plan and paste-ready prompts.
What was checked
Tie Zhu read the full post and article; Codex reviewed the handoff and overlap without independently reading X. Images not viewed, product details not checked line by line, and the workflow was not reproduced.
A single-job Bot checks selected Polymarket markets every 15 minutes and reports time, price, a short reason and rules after a 5% move; it never trades, and price movement is not resolution.
AppsPolymarketWhat was checked
Tie Zhu read the post and prompt; Codex reviewed the handoff without independently reading X. Demo video not watched; not reproduced. Financial risk applies, and the 15-minute cadence may use more allowance.
Keep a small Bot roster, save Skills while teaching, back them up to private GitHub, filter events before hourly Routines, script repeated steps, share Bots in Slack, and use voice-dump restatement.
AppsSlackGitHubWhat was checked
Tie Zhu read the post and the author's follow-up; Codex reviewed the handoff and deduplication without independently reading X. Template and media not opened; not reproduced; author-reported by a Grok Bot team member.
Run the morning check once and fix it, save it as a Skill, schedule weekdays at 7:30 with no sending or buying and explicit failed-site reporting, then use Test and review the output.
What was checked
Tie Zhu read the post; Codex reviewed the handoff without independently reading X; not reproduced.
Add one line to every Routine: end each report with Incomplete, Errors and Uncertain columns, making issues that once needed follow-up questions visible immediately.
What was checked
Tie Zhu read the post; Codex reviewed the handoff without independently reading X. The copy link and image were not opened; not reproduced.
A categorized directory of publicly accessible Grok Bot Pages with screenshots, creators, source links and copyable prompts, requiring real data or working controls and no private details.
What was checked
Read the README, rules and representative entries. Pages, controls, data and attribution were not individually verified, several examples use sample data, and listing is not a security or privacy audit.
Six question groups generate a manager, growth roles, shared rules, a review sheet, Routine plan and setup checklist, starting with one channel and human approval for publishing and spending.
AppsXBeatAPIBeatBrowserWhat was checked
Read the README and boundaries. No Skill, role, account or BeatAPI/BeatBrowser integration was installed or used. The vendor promotes optional services, and actual execution depends on the agent and tools.
A Bot launches a T3 worktree, then a Stop hook reads opt-in marker and status files and calls that Bot's Routine webhook, with duplicate suppression, retryable failures and local secret files.
AppsT3 CodeCodexClaude CodeWhat was checked
Read the README, state model and limits. No hook, config, MCP authorization, webhook secret or callback was used. The full-access recommendation expands permissions; scripts and T3 compatibility were not audited.
A Python reference uses one Monday clock and ledger DONE events for seven Bots, stops at read caps, keeps merge and posting disabled, requires approval for every purchase, and briefs only on changes.
AppsXGitHubWhat was checked
Read the README, architecture and safety boundaries. Nothing was cloned or run, and no account or purchase was used. The fixed-week demo is scripted; adapters, quota figures and concurrency were not independently audited.
As of October 31, the directory lists 7,826 public templates from 3,801 creators, organized by job with schedules, apps, marketplace status and three access-risk levels.
What was checked
Read the README, legend and representative entries. The 7,826 templates, creators and access ratings were not individually verified or audited. This evolving index differs from the smaller snapshot and Shelf; counts change quickly.
Four role packs cover profit reconciliation, product drafts, creator outreach and multi-store reporting, with identity checks, publishing, pricing, commissions and payments behind human approval.
AppsDouyinFeishuTaobaoPinduoduoWhat was checked
Read the role, installation and data-boundary sections. No shared Bot, shell installer, Skill, Douyin login or order data was used. Scripts, platform rules and backend compatibility were not audited.
A daily Bot ranks ten niche posts by outperformance, flags 3x outliers, records links, metrics, hooks, reasons, sentiment and takeaways, then proposes trends and ten ideas using visible data only.
AppsXElevenCreativeClaudeWhat was checked
Tie Zhu read the article and prompts; Codex reviewed the handoff and deduplication without independently reading X. Media not viewed; not reproduced. The vendor's later product and Claude workflow are excluded.
A scheduled scout builds a six-field topic log; a lead coordinates research, writing, design and publishing Bots that place long posts into X drafts for human review.
AppsXNotionFeishuWhat was checked
Tie Zhu read the full X article; Codex reviewed the handoff and deduplication without independently reading X. Images not viewed and workflow not reproduced. Quota, pricing and the third-party Skill remain unchecked.
A one-line prompt asks the Bot to list what must be true for a product idea, sourced evidence for and against it, and the cheapest test.
What was checked
Tie Zhu read the post and the prompt in the author's first reply; Codex reviewed the handoff and deduplication without independently reading X. Image not viewed; not reproduced.
Provide a VIN or year, make, model and mileage; the Bot tracks service, reminds when due, and lists DIY parts or a shop, without ordering, booking or paying.
What was checked
Tie Zhu read the post and prompt reply; Codex reviewed the handoff and deduplication without independently reading X. Image not viewed; not reproduced. Check schedules against the manufacturer manual.
The prompt inventories Bots and apps, checks the plugin directory for overlap and sensitive access, scores fit, time and risk, keeps the top six, flags idle plugins and plans a real test for each.
What was checked
Tie Zhu read the full prompt; Codex reviewed the handoff and deduplication without independently reading X. The linked breakdown was not read and nothing was reproduced. Each authorization still needs review.
Turn a tested Skill into a Monday 9am Routine, report only key points and sources, change or share nothing, verify the schedule and time zone, and test with safe inputs first.
What was checked
Tie Zhu read the full X article; Codex reviewed the handoff and deduplication without independently reading X. Image not viewed, not reproduced, and current UI paths were not checked.
A Chinese guide builds from one Bot to a managed team, uses a six-part work brief, saves successful workflows as Skills and Routines, and provides research, content and engineering playbooks with approval gates.
AppsGitHubSlackCursorXWhat was checked
Read the complete README and public examples. No template, team or playbook was run. Version 1.0.0 dates to 2026-09-09, so plans, UI and screenshots may be stale.
The repository documents shared-computer state with architecture, tool catalogs, process maps, session and action traces, JSON schemas, collectors and verification, with explicit secret redaction before Git sync.
AppsGitHubWhat was checked
Statically read the README and architecture docs. No collector, verifier, daemon, shared-computer read or Git sync was run. This is an author setup, not official telemetry, and privacy claims are unaudited.
A reviewable workflow combines scripts, storyboards, brand kits, captions, music and templates for 16:9, 9:16 and 1:1 renders, with approval before paid services, publishing or irreversible actions.
AppsYouTubeRemotionHyperFramesWhat was checked
Statically read the README and two Skill entry files. Nothing was cloned, installed, rendered, paid for or uploaded. Output quality, media rights and third-party costs were not independently verified.
The author reports Bot-driven pull requests for a voice-astrology site, deployment, database, email and Stripe, plus multilingual videos and a 9am operations Routine, with human publishing and payment approval.
AppsVercelStripeYouTubeTikTokInstagramWhat was checked
Read the post and author replies. Claims include 15–20 minutes daily, 7–9 visitors, 14 accounts and zero sales, but no code, billing, analytics or security evidence was available. Nothing was reproduced.
When no connector exists, use the shared browser with human login takeover, define file inputs and outputs, capture repeatable steps with Teach a task, and require approval plus backups for risky changes.
AppsGoogle DriveDropboxToastSquareMindbodyVagaroJobberHoneyBookPlanning CenterZendeskWhat was checked
Read the full guide. No shared computer, sign-in, upload, local execution or Teach a task recording was used. Third-party app compatibility and terms were not verified.
Define a Routine by name, trigger, sources, steps, output, boundaries and failure handling; test narrowly first, then start with briefings, pull-request risk summaries or draft-only inbox triage.
AppsSlackGitHubGmailGoogle CalendarNotionGoogle DocsWhat was checked
Read the full guide and sources. No Routine or connector was created. The page reflects 2026-10-26 and app 0.75.8; UI and numeric limits are time-sensitive.
Write Skills with purpose, inputs, steps, outputs, boundaries and examples; save only after a successful run, split large workflows, use Teach a task, version and prune, and review before team sharing.
AppsXWhat was checked
Read the full guide and source list. No Skill was created, tested or published. UI paths, app version 0.75.8 and quantity claims are time-sensitive and were not reproduced.
Tag @bot with a clear task on X to hand the post to your primary Bot. The public reply only confirms receipt; quoted media and regional availability may be limited, and the X plugin is optional.
AppsXSlackWhat was checked
Read the full official page. No X post, @bot tag, X connector or Slack action was used. Regional and media support are time-sensitive.
Govern Shared Bot access and auto-add, share Pages deliberately, deploy Slack apps, manage secrets through Vault or 1Password, keep Agent Email draft-first, and use admin analytics for adoption.
AppsSlack1PasswordGmailOutlookWhat was checked
Read the complete official page, updated 2026-10-28. No team, Shared Bot, Slack app, Agent Email, analytics report or permission change was created.
Define a role by job, sources, outputs and boundaries; validate one safe task, save the successful workflow as a Skill, schedule it as a Routine, and keep external actions behind approval.
AppsGitHubSlackGmailWhat was checked
Read the complete official page, updated 2026-10-28. No Bot, Skill, Routine or connector was created; availability can vary by plan and region.
Use Vault for logins, cards, contacts and secure forms; inject secrets through Vault or 1Password instead of chat, allow OTP retrieval when intended, and keep payment use behind approval.
Apps1PasswordGmailOutlookLinkX MoneyWhat was checked
Read the full official page. No credentials, cards, contacts, email or payment accounts were added, and no payment was attempted. Availability is time-sensitive.
Thirty routine ideas span inboxes, Slack, calendars, research, finance, site checks, dependencies, pull requests and travel, with every example stopping before sending, posting, paying or upgrading.
AppsGmailSlackGoogle CalendarXGitHubWhat was checked
Read the full list and FAQ and checked official scheduling guidance; none of the routines were run. The page says it works as of 2026-10-20; linked prompts and the paid pack were not reviewed.
Store the learner profile in a file, verify each canonical page, correct labels, durations and prerequisites, save a narrow Skill, retest it on a new goal, then schedule a weekly Routine that stops on missing sources.
AppsDataCampWhat was checked
Read the key steps, two correction rounds and final routine result; images and embedded video were not fully inspected and nothing was reproduced. Plan details predate the October subscription change.
A Skill sends one idempotency-aware webhook request to a dedicated Bot and collects its outbox result later, with explicit started, timed-out and unknown states plus strict secret and approval rules.
AppsGitHubSlackGmailGoogle CalendarGoogle DriveWhat was checked
Read the complete Skill and security notes and checked official webhook semantics; nothing was downloaded or run. Always-allow local execution is account-wide and creates a serious secret and prompt-injection risk.
A job-grouped directory of public x.ai/bot links, with clear notes on what a template copies, what it does not copy, and why every third-party template should be inspected before adding.
What was checked
Read the README, categories and representative entries; no template was opened, imported or audited. This is a 2026-09-03 snapshot with one commit, not an official ranking.
Pick one small recurring task, run and correct it once, save and retest it as a Skill, schedule it as a Routine, keep sends, purchases and deletes behind approval, and add only one new task each week.
AppsGmailWhat was checked
Read the full thread and comment. The author discloses AI-assisted drafting and human editing but gives no test record. An unrelated product-future claim is excluded and unverified.
Prompt: compare four homeowners-insurance quotes against an attached policy, draft requests to three local contractors, and send or buy nothing without approval.
What was checked
The post and the author's first reply were read; the image was not viewed and nothing was reproduced. Quote availability depends on insurer websites.
Ask for a page with the three numbers you check each Monday, name each source, set a routine to refresh it an hour before viewing, and keep sharing private.
What was checked
The complete text-only post was read; nothing was reproduced. The Page feature is recent, so use the current official UI as the source of truth.
When Bot, Cloud Agents and Origin degrade together, the whole bot-to-cloud-agent-to-repo chain fails; add an on-failure stop rule instead of blind retries.
AppsCursorWhat was checked
The short post text was read; linked media was not opened and nothing was reproduced. A promotional expansion is merged but not attributed to the original author.
One director, three leads, a silent 2–5am night shift and one 6:30 morning package replace direct reports from 20+ Bots, while posts, sends and payments stay behind approval.
What was checked
The full post was read; the image and quoted video were not viewed and nothing was reproduced. It provides no exact prompts or routine configuration.
Open the Cybertruck browser, visit x.com or grok.com, choose the Bot, tap voice call and allow the Tesla microphone, with no extra app required.
AppsTeslaWhat was checked
The post text was read; the image was not viewed and nothing was reproduced. This is one author's Cybertruck report; other vehicles and regions are unknown.
Build from scratch, connect only needed apps, practice with Drive and Gmail, make routines with no-send rules, group Bots under an admin Bot, save skills, and keep outbound approval.
AppsGmailGoogle CalendarGoogle DriveWhat was checked
Most article text was read; routine details were skimmed, and screenshots and embedded posts were not viewed. Pricing is time-sensitive and was not reproduced.
Connect via ContentStudio MCP or CLI plus Skill, then run six draft-only social routines for scheduling, reports, inboxes, approval reminders, gap checks and new media.
AppsContentStudioWhat was checked
The full article was read, but images were not viewed and nothing was reproduced. The third-party MCP, CLI and skills were not installed or audited; the article is vendor-adjacent.
Connect X and verify with a read-only prompt, start from templates after checking permissions, clean memory every two weeks, back up skills daily, and cap on-demand usage.
AppsXNotionGitHubCursorBitwardenWhat was checked
The full article and four text tables were read; images and the quoted video were not viewed, and nothing was reproduced. It contains a referral link and unverified third-party tools.
A Traditional Chinese walkthrough from plans and the first Bot to app connections, approval rules, routines, skills and Teach a task, with a practical inbox prompt.
AppsGmailGoogle CalendarGoogle DriveSlackNotionGitHub1PasswordWhat was checked
Read the full article and cross-checked current official setup, routine and approval docs. Images were not all inspected and nothing was reproduced. Pricing is time-sensitive as of 2026-10-29.
A seven-path guide to waking Grok Bot, including a Cloudflare Worker that verifies and relays LINE events. It separates accepted webhooks from completed work and documents author-tested limits.
AppsLINECloudflare WorkersGitHubSlackWhat was checked
Read the full article and code excerpts, cross-checked against official routine/webhook docs. No bridge deployment or account connections. GitHub, LINE Reply API and failure behavior remain author-reported; the article links the author's automation services.
A Stream Deck key opens a selected Bot and uses a local helper to press the voice controls. The author says it makes no network requests, but it needs macOS Accessibility, an English UI and local Bot data.
AppsStream DeckmacOSWhat was checked
Read the README, repository tree and v0.2.0 release metadata. The repository contains docs, images and a license; the plugin is a release binary. No download, install, Accessibility grant, binary audit, network test, signing/notarization or compatibility verification.
A community MCP plugin and six Skills read Tesla Fleet data. Commands stay hidden by default, require preview and per-call confirmation when enabled, and are forbidden from routines, webhooks or other agents.
AppsTesla Fleet APIGrok BuildCursorWhat was checked
Read the README, boundary and setup Skills, file list, and official Tesla billing/command docs. No install, vehicle authorization, API call or code audit; four Skills and experimental commands were not fully reviewed. Policy text is not an enforced security boundary or endorsement.
A community directory groups shared Bot templates by category and creator, with search, source posts, optional code links and creator-supplied connector, routine and permission metadata. Coverage varies by entry.
What was checked
Read the README and statically counted bots.json: 652 unique template URLs, 467 with source posts and 625 with creator pages. Templates, attribution, permission labels, votes and moderation were not individually verified; creator-supplied metadata is not a security audit.
A Monday routine reads the prior week's bill and receipt emails, lists amounts due within 14 days and checks named charges, while keeping the report short when nothing needs attention.
What was checked
Read the full post and visible comments. The 30-email result is author-reported; no prompt, mail provider, connector permissions or output was available, and no inbox was accessed. The author promotes a newsletter in a comment. This records the narrow design, not a reproduced result.
The Integrator template centralizes account, plugin and customization questions, and claims to pause preset routines and check schedule or Skill conflicts before a newly imported Bot is put into service.
AppsGrok Bot MarketplaceGmailGoogle CalendarWhat was checked
Read the Reddit post and public template title/description; the template page displays the byline “Your.” No template import, hidden-instruction access, Gmail/Calendar connection, or verification of automatic discovery, routine pausing or account selection. All capability claims remain author-reported.
A Skill checklist for unattended jobs: specify timezone and outputs, handle missing inputs, avoid duplicate actions, and leave approval-dependent work as drafts. Triggers use narrow sources and deduplication.
What was checked
Read the root README and this full Skill. No installation, code execution or routine testing. Approval expiry and UI details remain unverified; frequency thresholds are heuristics, not product limits.
Ten Grok Bot Skills cover triage and routine decisions. The routine-wake example compares local state first, requests external advice for ambiguous cases, and retains mandatory local alerts and failure rules.
AppsJevTypeSafeWhat was checked
Read the README and full routine-wake Skill; only catalog descriptions for the other nine. The repo credits Steve Darlow and hermes-jev-skills. No external calls, install, cost, quality or privacy tests; instructions are not enforced controls.
A Feishu long-connection bridge checks allowlists, invokes a Grok Bot Routine webhook, and returns outbox progress and results by task ID. The docs cover retries, single-instance operation and attachments.
AppsFeishuWhat was checked
Read the README, security and protocol docs, without running or auditing code. 2xx means accepted; retries may duplicate runs. Same-machine files are not isolation. Outbox attachments may reference any readable absolute path; mention checks depend on botOpenID.
The author dispatches prompts from Grok Bot through Herdr to Pi, then returns completion webhooks via a custom plugin. Follow-ups describe remote Windows access and machine/session labels for routing.
AppsHerdrPiTailscaleWhat was checked
Read the original post and author follow-ups. Herdr dispatch, a custom Pi completion plugin, Tailscale and session labels are described. Plugin source and full setup are unavailable; no reproduction or herdsman compatibility verification.
The author demonstrates a repeatable restaurant slideshow workflow, then uses Grok Bot to gather location and image material and assemble TikTok slides. Visual preferences are taught through examples.
AppsTikTokGoogle SearchPinterestWhat was checked
Read the full original post, without viewing images or reproducing it. It supports research and slideshow creation, not verified automatic uploading. Time, views, costs and image rights are unverified; exact publication date is unknown.
Use a shared Google document for Grok Bot and Muse handoffs, labeling time, author, open questions and completion. Each assistant reads outstanding items and appends updates; the author excludes secrets.
AppsMuseGoogle DocsGoogle DriveWhat was checked
Read all five steps. Shared Drive access is the author’s prerequisite, not access granted or tested here. No private documents, permission changes or reproduction; reported efficiency is unverified. Distinct from a document-indexing workflow.
Assign one owner to each area and route outputs through a chief-of-staff Bot, staying quiet when nothing changes. An editor compiles weekday columns; the author favors reusable templates and a small team.
What was checked
Read the original post. Exact prompts and template links are not released, so this is a configuration case, not an installable template. No team or digest testing; merged retellings are based on the handoff.
Extract a Skill from a successful comparison task, separate lasting preferences from run inputs, and retest with new products, missing budgets and interrupted runs before proposing routines or shared templates.
What was checked
Read all ten sections and six prompts, plus official Skill docs. The article’s slash-menu instruction differs from current name-based invocation. No execution, images, quoted video or template verification; this is distinct from monthly audits.
Split SEO work into coordinator, analyst, writer and refresher roles, using ChatSEO MCP data and a status board. Schedule daily or weekly work, with human approval for publishing, site changes and redirects.
AppsChatSEOGoogle Search ConsoleWordPressWhat was checked
Read the article and three prompts. ChatSEO is the author’s commercial service; plans and MCP access were not independently verified. Images were not viewed and are not proof of this Bot setup. No account connections, authorization or runtime tests.
Six workflows and fourteen prompts cover content funnels, leads, sponsors, reusable service setups, skill interviews and job preparation. Several stop at drafts before outreach or application submission.
AppsXWhat was checked
Read the full article and fourteen prompts in the normal browser after a web-fetch 403. No image, template, earnings or runtime verification. Shared-memory assumptions are not treated as universal facts. Merged promotional posts rely on Tiezhu’s handoff.
An independent client groups and pins Bots, searches chats and opens computers, with a desktop-helper and local iPhone pairing guide. It uses an existing account; version and device coverage remain limited.
AppsGrokDeckWhat was checked
Read the full README. The author says its helper reads OS-protected account storage and holds decrypted session credentials; this was not audited. No download, install, pairing or user-cache access. Mac/iPhone checks and incomplete Windows testing are author reports, not our tests.
Nine .me Bot page layouts use structured profile, Bot-card and link fields while preserving layout code. Billboard also has a Vercel static export. Reviewed as one template collection.
AppsVercelWhat was checked
Read the full README and checked the Stack source for its grok/canvas import, layout identifier and placeholders. No rendering, Bot install or Vercel deploy. Other layouts, analytics and security were not audited. Publication date is unknown.
Archive Bot digests as dated Markdown with fixed metadata, render category pages in VitePress, and deploy through GitHub Actions to GitHub Pages. The docs specify paths and deployment settings.
AppsGitHubVitePressGitHub ActionsGitHub PagesVercelWhat was checked
Read the root, content and web READMEs in full. Paths, metadata, URL mapping and deployment are documented. No code, Bot submission, access or production deployment was tested. The author’s direct-main policy is not authority for this site; publication date is unknown.
A route map compares webhook templates, desktop CLIs, reverse handoffs and shared boards, linking their source projects. Some BDK commands are outdated; this is a reference, not a current installation guide.
AppsChatGPTClaude CodeCursorCodexCrosswalkWhat was checked
Read all eight sections, two bridge READMEs, Grok Bridge alpha and Crosswalk docs. Official docs renamed @cursor/bdk to @grok/agent-sdk; grokbot-agents and attachMCP claims remain unverified. Reference only: no install, audit or compatibility test.
Six prompts cover build plans, launch radar, source packs, client deadlines, inbox triage and weekly reviews, with draft-only actions, explicit missing-source handling and quiet runs when nothing changes.
What was checked
Tiezhu and Codex both read the full X article and six prompt blocks. The source pack is an on-demand skill. Quota, usage-pool and no-dry-run claims were not verified or adopted. Cover unseen; not reproduced. Distinct from the author’s service-delivery guide.
Diagrams cover shared cloud computers, local access, memory, approvals, drafts, skills, routines and staged handoffs. Key frames were checked against official docs; no full audio review or hands-on test.
AppsSlackGitHubWhat was checked
Read the post and sampled key frames from the 6:53 explainer, cross-checking the cited official docs. No text-track captions, successful media download, frame-by-frame/full audio review or reproduction. Usage illustrations and integration coverage are not guarantees.
The author joins a Bot to a tailnet and routes coding to a Mac via T3 Code MCP and OAuth. T3 docs require HTTPS for remote sign-in; read-only connection access cannot start threads. Not reproduced.
AppsT3 CodeTailscaleClaude CodeCodexWhat was checked
Read the full eight-step post, not its video, plus T3’s Outside agents doc. HTTPS, OAuth and connection scope were checked; a read-only task is distinct from read-only connection access. No install or authorization. Date follows the visible Oct 27 label, correcting the collector’s Oct 26 without guessing time zones.
The guide audits recurring charges with Finance/Plaid, then adds a dashboard, weekly digest and bill reminders without cancelling services. The official listing confirms read-only access and stored account data.
AppsFinancePlaidRemindersWhat was checked
Tiezhu and Codex read the article. Codex read Finance’s full public marketplace permission text: read-only Plaid access, synced/stored account data and unlinking. No account connection, financial-data access or actions. Pages/phone Reminders, images and results remain untested. Date corrected to the visible Oct 26 label.
One episode checks recipients, subject, body and attachments before sending, stopping for placeholders or missing items. This is the author’s checklist, not a verified technical guarantee or review of the whole series.
What was checked
Read the post and sampled the 1:11 clip at about 0:23, 0:33, 0:47 and its ending. No text-track captions or full audiovisual review. The author labels the voice as their AI clone. Linked site returned 404; no send test, guard audit or other-episode review.
An MCP bridge exposes workspace tools via local stdio or remote OAuth. Its optional agent mailbox only queues messages; a Grok Bot parent routine must deliver work and record replies.
AppsChatGPTClaude DesktopCursorGitHubCloudflareWhat was checked
Codex read README use cases, tools, transports, the reported session, mailbox behavior and known limits. No install, doctor, tunnel, workspace grant or OAuth audit. Linked PR not independently inspected. Queuing does not itself wake a Bot.
A local dashboard logs weekly usage percentages and work check-ins, with history, search and pace estimates. Optional sampling reads Grok Build’s usage screen; forecasts and Bot quota mapping are untested.
AppsGrok BuildWhat was checked
Codex read relevant README sections and recorder/reader functions, merging the Reddit announcement and repository translation. No full audit or execution. Linux preferred; macOS untested, Windows auto-read experimental. Optional GitHub update checks use the network; quota mapping unverified.
A six-file business handbook supports six Bots for briefings, support, review, content, research and finance, with limited access, human approval and weekly routines.
What was checked
Tiezhu reports reading the full article and one code block; Codex reviewed the handoff and attribution. Images unseen; no Bots created or plan reproduced. Price, quota and outcome claims omitted.
A parent reports four subject Bots and a coordinator using a local SQLite question bank and Notion practice logs. A role-design case; the connection to the local bank is not documented.
AppsNotionSQLiteClaudeWhat was checked
Tiezhu reports reading all six posts; Codex reviewed the handoff. Media unseen, question-bank code unavailable and cloud-to-Mac access unexplained. Proposed only as an author-reported case, not a reproducible tutorial or outcome proof.
A one-job Bot triages messages, checks during the day, prepares a morning brief and drafts Slack-triggered replies. Corrections inform memory; sending requires human approval.
AppsGmailSlackTelegramComposioWhat was checked
Tiezhu’s Monday retry reports reading the article and five prompts; Codex checked the handoff, not inboxes or DMs. An editing placeholder remains, examples are unseen images, and Composio/Telegram permissions are unverified. No routines tested.
A Chinese guide compares Marketplace and Use Cases, then recommends checking connector dependencies and setup needs before choosing a simpler template and trying a small task.
What was checked
Tiezhu reports reading the Chinese article. Codex separately read the official template guide, not the X body. No catalog-wide check or installation. Author counts, regional availability, pricing and timeline claims omitted.
Seven author-made Bot templates cover local deals, reminders, websites, clips and Bot audits. Names, authors and public previews were checked; full configuration and permissions were not inspected.
AppsYouTubeWhat was checked
Read the X article, author directory and all seven x.ai previews; names, authors and purposes match. Public pages expose summaries, not full instructions, skills or permissions. No Add click, terms acceptance or install. Mail/credit access, repairs and video results remain untested.
Six studio roles coordinate app work. Builder prepares briefs for a coding agent; handoffs need evidence, checks run on routines, and spending, submission and release require approval.
AppsSentryApp Store ConnectGoogle Play ConsoleWhat was checked
Tiezhu reports reading the article and 21 code blocks; Codex checked the handoff. Proposed Sentry/store integrations were not connected. Builder drafts briefs; the unrelated, unwatched video’s cost claims are omitted.
A chief-of-staff interview defines voice, audience and content pillars before narrow research, writing, checking and tracking roles use X and Notion, with human review before posting.
AppsXNotionWhat was checked
Tiezhu reports reading the article and 18 code blocks; Codex checked the handoff and duplicates. Images and metrics screenshots unseen. No X/Notion connection, team setup or usage benefit tested.
Five Bot roles use Treg data to find content gaps, draft with Claude Code, and queue outreach and X replies for approval. A commercial-service workflow with untested results.
AppsTregClaude CodeXChatGPTClaudePerplexityGoogle AI ModeWhat was checked
Read the full X article and six code blocks: one setup block plus five roles. Checked Treg’s setup, team/billing docs and public visibility catalog, not live calls. No install, login, secret upload or tool execution. Author ties, provider coverage, hidden searches, costs and gains remain unverified.
A versioned shared folder routes detailed rules, uses a single writer, records progress and evidence in TODO files, and reviews Bot feedback weekly before changing policy.
What was checked
Tiezhu reports reading the article and four structure/template blocks; Codex reviewed the handoff. Four scripts are only described, not inspected or run. Media and the earlier article unread; external rules are source data.
The author binds a Bot voice action in Shortcuts to a custom Vocal Shortcut, with a possible extra tap when the phone is locked.
AppsShortcutsWhat was checked
Tiezhu reports reading the post; Codex checked the handoff and Apple’s general Vocal Shortcuts instructions. No phone test of Bot actions, locking or device/iOS support; media unseen. Distinct from Back Tap.
Define roles, inputs and output locations, then use briefings and a chief of staff to coordinate research, scripts, packaging and Shorts. Full auto-captions and description read; no visual review or reproduction.
AppsYouTubeXHiggsfieldWhat was checked
Followed the X promo to the original 21:04 episode; read full English auto-captions and description, not video images. Names follow the description. The separate-computer claim is not adopted: official docs specify a shared account computer. Group/UI limits are dated observations; no templates requested.
A task bridge queues read-only work for Grok Bot, tracking leases, idempotency keys and structured results. The author case returns evidence through a manually triggered bot conversation.
AppsCodexClaude CodeCursorCloudflare WorkersCloudflare D1Cloudflare R2What was checked
Read the author V2EX post, both READMEs and the September 13 case. Author-reported; raw logs, code and auth not audited. Unattended routines remain unverified. No install, deployment or X-link access.
A receiver-side skill caps senders, quiet hours and wake rates, documents a Grok Bot routine-webhook adapter, and distinguishes accepted wake-ups from completed work.
AppsTailscaleCodexWhat was checked
Read README wake/status sections and the full wake skill. No install, receiver, Tailscale or webhook call. Grok adapter, authentication and optional Codex queue compatibility not tested; messages are not authorization.
A Chinese playbook defines supervisor-specialist teams for eight roles, with seven-part bot descriptions and handoffs that report file paths, acceptance checks and unresolved issues.
AppsGmailCalendarSlackGitHubXNotionWhat was checked
Read both READMEs and the full handoff template, merged as one work. All 35 prompts, skill loading and eight teams not tested. Named tools are examples; product-architecture claims not independently verified.
A bridge checks Telegram updates and allowlists before waking a bot routine. Its setup skill walks through configuration; replies use short-lived tokens scoped to a chat and topic.
AppsTelegramCloudflare WorkersVercelWhat was checked
Read README architecture, hosting, routine and reply sections plus setup-skill access and reply rules. No code audit, installation, bot creation, secrets or messages; pricing and compatibility unverified.
A persistent forwarder batches WeChat messages into a bot webhook, stops superseded replies and separates delivered messages from queued sends that wait for a later conversation window.
AppsWeChat ClawBotWhat was checked
Read README architecture, payload and setup sections and the inbound template, an equivalent example rather than an exported live routine. No code/terms audit, binding, WeChat access or tests; version/window claims are author reports.
A public plugin example long-polls Telegram and wakes a bot through an authenticated webhook. Its skill separates listening, typing, sending and checking the send result.
AppsTelegramWhat was checked
Read the README and full gateway skill. No install, group-privacy change or token access. Code, private-path/secret-flow compatibility and marketplace listing unverified; timing and native-channel claims not adopted.
A Chinese directory combines ecosystem links and original playbooks. Its war-room example coordinates focused roles, task boards and file handoffs, with human review before publication.
AppsClaude CodeWhat was checked
Read README guidance and the full war-room example, whose products and outputs are fictional. Directory entries and execution not independently tested; author badges are not our tests. Translations merged; no X-index scan.
An author organizes Notion setup around connection state, workspace labels, read-only trials and recovery checks, validating reads before trying edits in a dedicated page.
AppsNotionWhat was checked
Read the public guide, updated October 21. The author sells a template pack; none acquired. UI, access scope and retention claims not individually verified; no Notion connection or edits.
A prompt organizes supplied or authorized statements and email receipts into recurring charges, monthly totals and items for human review; amounts and cancellation behavior are unverified.
AppsEmailWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full post incl. prompt; image not viewed. Requires bank/email access; figures author-reported. Not reproduced. Bookkeeping organization only; no user bank/email data read and no subscription cancelled.
An author presents four stages: validate one task, refine the workflow, add triggers and build a team, alongside an unofficial Chinese learning site and template directory.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full X article; the site, repo and templates were not opened. Author promotes their own site. Not reproduced. This entry covers the article, not verified site chapters or templates. Site counts and old plan claims are not adopted.
Start with one bot and one verifiable task: files, scoped research and read-only connected apps. Define approval boundaries before saving skills, scheduling routines or adding roles.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full X article; the author's 10-21 promo post is merged (video not watched, user figures unverified). Not reproduced. The September 11 original was rediscovered through an October 21 promotion; it is not a new release.
A weekday prompt searches competitor updates and relevant purchase requests, returning a linked brief in separate sections without replying or interacting.
AppsXWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the post and the prompt in the author's first reply; image not viewed. Not reproduced. Only the collector-read prompt is reviewed; no test scan of X or personal-account connection occurred.
An author combines local MCP, routine webhooks and Tailscale Funnel for two-way Claude Code handoffs, with a setup prompt and a rule that received replies are information, not instructions.
AppsClaude CodeTailscale FunnelWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full post incl. prompt; image not viewed. Risk: exposes a local server publicly; manage tokens. Not reproduced. The referenced older four-API-call hold was unavailable locally; no completed comparison with it is claimed. This is a separate original-post candidate, not an audited public endpoint or integration.
The author proposes forwarding domain aliases into a bot inbox and routing support and personal mail by recipient address, with separate authority for setup and external replies.
AppsCloudflareEmailWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full short post; a Chinese paraphrase by vikingmute is merged here. Requires granting domain account access. Not reproduced. Current pricing, preserved headers and actual routing behavior were not verified.
An author separates site monitoring, media research, AI news and reminders into four bots. Unspecified plugins and backlink automation are not treated as reusable instructions.
AppsXXiaohongshuDouyinWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full post; image not viewed; the backlink plugin is unnamed. Not reproduced. Xiaohongshu/Douyin access methods and quota impressions were not verified or generalized.
Before outreach, search authorized CRM and inbox connections for existing relationships at a target company, rank them and draft an introduction request without sending.
AppsCRMEmailWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full post including the prompt; image not viewed. Requires CRM and email connections. Not reproduced.
A prompt turns an earnings call into a newspaper-style page with direct quotes and accessible source links. This entry describes the artifact design, not verified financial conclusions.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full post incl. prompt; generated page and image not opened. Not reproduced.
Ask each routine to justify its frequency, then propose changes based on events, update speed and deliverables. Review the reasons and plan before approving edits.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full post incl. prompt; video not watched. Cost figures are author-reported. Not reproduced. Savings and fixed hourly/five-minute intervals are not general guidance. Unlike existing trigger overviews, this entry focuses on a schedule-audit prompt.
The author describes handing a shopping task from in-car Grok to a bot and organizing it in Apple Notes; messaging/notes integration and version or region support remain unverified.
AppsTesla GrokApple NotesWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full post; video not watched. Availability by region/version unverified; integration details not given. Not reproduced. An author-reported use case, not a general setup tutorial or official feature announcement.
Two accounts share a self-hosted fitness dashboard through role planning, multi-user changes, template handoffs and bot-email synchronization, as described by the author.
AppsEmailWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the full post. Involves a self-hosted server exposed publicly; credentials handed over by the author. Not reproduced. A collaboration case only; templates, dashboard access isolation and fitness advice were not verified.
The author describes routing cloud-computer traffic through the local machine for connectivity issues; the option name, location and applicability need current-version checks.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication only, without independently reading X or testing the workflow. Read the post text; screenshot not viewed. Setting name per author, may change by version. Not reproduced.
Official guidance covers directed messages, group chats, asynchronous handoffs and threads, alongside editable drafts, pending approvals and redirecting work in progress.
AppsSlackWhat was checked
Read the official page, updated October 17. Reviewed groups, handoffs and drafts; no group created, message sent or account changed. Group chats do not expose every approval or login card.
Official guidance covers attachments, file outputs and conversion plus Pages for reports, dashboards and trackers stored in Library and updated by Routines; public links require careful privacy review.
AppsGoogle DocsGoogle SheetsGoogle SlidesGoogle DrawingsWhat was checked
Read the current official page, including Pages, Library, sharing and Routine updates. No file, Page, Routine or share link was created; feature requirements are time-sensitive.
The author description outlines integrations, permissions, routines and bot chains, with examples of organizing public information in Notion and preparing work for Cursor cloud agents.
AppsGmailNotionCursorSlackGitHubWhat was checked
Read the full public author description and metadata via ordinary HTTPS after a web-tool throttle, with TLS checks enabled. An August 25 backfill; video not watched, workflow not tested and no X user data collected.
Authorize and label each Microsoft connection, name the account in requests, and verify the mailbox behind each alias before using drafts or calendars. Work accounts may require administrator approval.
AppsOutlookMicrosoft 365ComposioWhat was checked
Read the vendor’s October 8 guide by Varshisth. Commercial source; no plugin install, Microsoft authorization, mailbox access or API call. Setup time, shared mailboxes and compatibility were not tested.
Use goal, tools, context, boundaries and acceptance criteria in task briefs, with role, skill, routine and review templates. Named apps are examples in the files, not tested integrations.
AppsSalesforceSlackLinkedInGmailWhat was checked
Read the role, brief, routine and safety sections plus prompt-templates.md. Only template methods are recommended; old UI, usage and product-triage claims are not adopted. No installation or app connection.
A community skill adapts drafts to context and owner rewrites, keeps drafts in the intended Slack destination, and waits for an explicit send instruction. It depends on the assistant’s Slack integration.
AppsSlackWhat was checked
Read the README and full slack-voice file. The README says Grok loaders are untested. This is a readable instruction template, with no loading or native-draft guarantee. No install or Slack access occurred.
The author’s macOS MCP bridge uses the accessibility UI to list bots, read chats, send messages and handle choices from coding agents, with an accompanying collaboration skill.
AppsClaude CodeCodexCursorVS CodeWhat was checked
Read the full README. No source-code audit, accessibility grant, MCP install or doctor run. Compatibility and implementation claims such as network or telemetry behavior were not tested.
A community directory groups skills, plugins and MCP projects with format and stability notes. Use it to find originals; directory labels do not constitute a security audit or site verification.
What was checked
Traced the discovered fork to its upstream and read extension, catalog and risk sections. Listed projects were not individually checked. Old pricing/platform claims and blanket verification claims are not adopted.
A Notion status field and role-specific columns coordinate checking, drafting and human review. Routines stop on empty results and cap batch sizes, with example skills and prompts.
AppsNotionWhat was checked
Read the article and two templates, published October 18 and revised October 19. State handoffs are not a locking guarantee. Author testing is not our reproduction; connector/UI claims and images were not verified.
An author separates bot orchestration, an Oracle-hosted MCPHub, compilation and local creative software, using SSH tunnels to reach Mac Blender, ComfyUI and emulators and hand off artifacts.
AppsClaude CodeOracle CloudMCPHubBlenderComfyUIHome AssistantAndroidWhat was checked
Read the October 19 original’s architecture, tool and artifact-handoff text. Images and linked implementations were not audited; no remote-control/MCP installation or reproduction. Account and model claims are not general guidance.
A nightly backup proposal covers settings, routines, memories and redacted transcripts, excluding passwords, tokens and cookies and stopping when sensitive material is detected.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication, without independently reading X or testing the workflow. The collector read the full article; cover not viewed; the template and the optional Mac extract script were not opened or run. Lockout is author-reported; user-made template; not reproduced.
Identify the printer and official manual, back up settings, then seek approval for network printing and scan-to-folder changes. Device and local-computer access require separate approval.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication, without independently reading X or testing the workflow. The collector read the post and the full prompt in the author's reply; video not watched. Needs local-computer access on the printer's network, which carries local and LAN control risk; disabling firmware updates is a security trade-off. Not reproduced.
Brief the outcome, sources, limits, deliverable and stop point; label sources for citations, review research notes before the result, and distinguish evidence from proposed next steps.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication without independently reading X or reproducing the workflow. The collector read the article text; the 4-minute video was not watched. Based on official getting-started material with fictional sample data. Not reproduced.
Three prompts cover public topic posts, mentions to review and timeline summaries, with a routine for repeat briefs. Personal content requires the account owner’s connection and approval.
AppsXWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication, without independently reading X or testing the workflow. The collector read the thread root and 3 follow-ups; image not viewed. The free X API credits claim is the author's; mentions and timeline need your own X connection. Not reproduced.
Seven service-delivery ideas organize content, research and reporting work around skills, routines and human acceptance checks. Revenue outcomes are unverified.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication, without independently reading X or testing the workflow. The collector read the full article and 3 code blocks; cover not viewed. Income ideas are the author's views, not measured results; pricing and dates are cited and time-sensitive. Not reproduced.
Single Grain’s owner outlines coordination, bot improvement, pre-call preparation, outreach drafts, topics and video roles. This is an author-reported case, without the full underlying prompts or skills.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication without independently reading X or reproducing the workflow. The collector read the full post; image not viewed. Author-reported practice at his own company; prompts and skills not shared. A Chinese recap is merged here. Not reproduced.
Start with a read-only public-source brief, validate it manually, then save a skill and routine with approval and stop rules. Plan and usage advice needs current official confirmation.
AppsXWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication, without independently reading X or testing the workflow. The collector read the full article and its 2 prompt blocks; images not viewed. Plan prices are the author's 2026-10-19 reading of official pages and may change. Not reproduced.
Define coordinator and specialist roles and working-file locations, validate tasks before saving skills, and specify routine timing, inputs, outputs, approvals and failure handling.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication without independently reading X or reproducing the workflow. The collector read the full X article and its 3 code blocks; images not viewed. Feature limits quoted from docs may change. Not reproduced.
Inventory bot settings, routine prompts, skills and checked files, keep copies outside the account, record credential sources without values, and periodically validate backups and rehearse rebuilding.
AppsGitHubWhat was checked
Tie Zhu read the post; Codex reviewed the handoff against yesterday’s still-pending Jay Graves template. This is a separate manual export and restore-check article. The loss and cause are author reports; not reproduced.
An ESP32-S3 display routes transcribed speech to an existing bot through a computer and webhook, then returns captions and audio. The author supplies backup, Wi-Fi and troubleshooting prompts.
AppsCodexGitHubWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication, without independently reading X or testing the workflow. The collector read the full X article and all 15 code blocks returned by the API; images not viewed; the zongzhu-bot repo and firmware were not opened or flashed. Third-party firmware, an always-on local backend and webhook token handling need checking. Not reproduced.
Map measurable weekly outcomes to a minimal roster, validate skills, keep unattended routines at drafts or reports, narrow event triggers, and periodically review roles and rework.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication without independently reading X or reproducing the workflow. The collector read the full X article and all 12 prompt code blocks; images not viewed; Substack link not opened. Similar theme to the KanikaBK entry but independent. Not reproduced.
A bot-inbox routine checks time zones and the live calendar before proposing meeting slots, with explicit limits on scheduling emails it may send and actions that need approval.
AppsEmailCalendarWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication, without independently reading X or testing the workflow. The collector read the full prompt in the original long post; image not viewed. The author's 2026-10-19 quote repost is merged here. Letting the bot send scheduling emails is the author's own choice. Not reproduced.
An author organizes bots around focused roles, standing preferences, outcome briefs, source checks and sending approvals, with a first-week plan. Growth outcomes remain author reports.
AppsXGmailGitHubWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication without independently reading X or reproducing the workflow. The collector read the full X article; images not viewed. Author builds TypeAide and says a bot helped write it; the 13x reach figure is a small single-account sample. Not reproduced.
Per-contact webhooks and keys deliver notes to another person’s bot. Each note is shown to its owner and does not itself authorize sending, spending or task execution.
What was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication, without independently reading X or testing the workflow. The collector read the thread root and 2 follow-ups; the template was not opened, so contents and permissions are unchecked. Cross-account webhook messaging needs your own security review. Not reproduced.
An author forwards Gmail to the bot inbox, surfaces mail needing decisions without automatic replies, and checks the flow with a test email. The forwarding scope still needs explicit review.
AppsGmailWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication without independently reading X or reproducing the workflow. The collector read the full post; receipt screenshot not viewed. Forwarding all mail shares its contents with the Bot. Not reproduced.
A two-layer backup plan sends selected working files to private GitHub nightly and archives to Google Drive weekly, with secret checks and routine failure reports.
AppsGitHubGoogle DriveWhat was checked
Tie Zhu read X; Codex reviewed the handoff and deduplication, without independently reading X or testing the workflow. The collector read the full post; images not viewed. Reset incidents are author-reported; the backup setup was not reproduced. Private repo and Drive access need your own review.
An author binds a bot voice action in Shortcuts to iPhone Back Tap and describes an Action Button option. Action names, app versions and device support require checking on the actual phone.
AppsShortcutsWhat was checked
Tie Zhu read the post; Codex reviewed the handoff, not X itself. Images, shortcut actions, version and device requirements were not independently checked. Similar Action Button reposts are merged; no reproduction.
The official guide covers synced mobile chats, voice, sharing files, reviewing message drafts and computer handoffs, while distinguishing routine editing and testing that still need desktop.
AppsSlackWhat was checked
Read the official page, updated October 17. No mobile trial, routine edit, Slack connection or permission change. This does not verify community shortcut actions or version claims.
The official guide covers Team Bot setup, shared skills and plugins, team memory, private chats and Slack access, including consent for personal connectors in shared conversations.
AppsSlackWhat was checked
Read the official page, updated October 16, including chat, credential and connector distinctions. No Team Bot was created or published; no Slack connection or permission change was made.
The official page distinguishes existing subscriptions from unified-plan access, shared weekly usage and dashboard locations, and notes that completed unification is irreversible.
AppsCursorGrokXWhat was checked
Read the official page, updated October 12. No price promise or switching recommendation is made; no account checkout, subscription unification, cancellation or purchase was performed.
The video description outlines research, coding, writing, operations and coordinator roles, with Gmail, Calendar, Notion, AgentMail, scheduled briefs and draft-only sending rules.
AppsGmailGoogle CalendarNotionAgentMailWhat was checked
Read the full author description and metadata via ordinary HTTPS after a web-tool failure. An August 28 discovery, not a new release. Video and prompt pack were not reviewed; old platform/plan claims are not adopted.
The author’s course description uses four layers to cover email, ClickUp work logs, transcripts, Slack-triggered routines and template sharing. Review is limited to the full description and chapters.
AppsClickUpSlackComposioWhat was checked
Traced the derivative guide to the original video and read the full author description and metadata via HTTPS. Not watched or purchased. Published August 31 in Beijing time; an older discovery, without outcome claims.
An author-reported content workflow separates monitoring, topic selection, drafting and review among four bots, with human editorial decisions and explicit source checks.
AppsXWhat was checked
Read the original article, focusing on its content handoff example. Images and workflow not tested. The site sells recharge services; old plan, pricing, platform and performance claims are not adopted.
The vendor describes saving GTM methods as bot skills and scheduling scoring or briefing routines. Its list-builder previews a sample before an approved export.
AppsExploriumVibe ProspectingWhat was checked
Read the bot-specific README section and list-builder/SKILL.md. Requires a vendor MCP/account; full exports use a paid path. No installation, API call, contact lookup or compatibility test performed.
The public skill describes site/file publishing, private Drives and version-conflict checks, explicitly including a grok-bot client identifier for cross-agent delivery workflows.
Appshere.nowWhat was checked
Read SKILL.md as source data, not task instructions. No installation, credentials, login, upload or publishing. Bot import, server permissions, retention and price claims were not verified.
An unofficial bridge links existing agents with Python, SQLite, MCP Events and webhook routines, separating delivery, claim, reply and completion with duplicate-handling limits.
AppsChatGPT DotWhat was checked
Read the V2EX introduction, repository README and acceptance notes. Results are author records; raw logs, code and authentication were not independently tested or audited. No accounts connected or messages sent.
An original first-task template turns approved interview material into a summary, draft posts and structured actions, with source references, missing-value rules and human acceptance checks.
What was checked
Read the public page via ordinary HTTPS after a web-tool timeout, with certificate checks enabled. The author says the template is untested. Old UI and plan details are not current guidance; no service purchased.
The author describes reading new forum topics daily, tracking seen items, writing linked English summaries and publishing a GitHub Pages digest only when there is new material.
AppsGitHub PagesWhat was checked
Read the original post and linked public site. Its visible October 16 page does not prove ongoing daily execution. No backend, forum access method or automation tested; no publication date inferred from relative time.
A bot role template structures issue clarification, investigation, tests, implementation and PR handoff, keeping merges and deployment subject to human approval.
AppsGitHubWhat was checked
Read the October 18 article and the publicly served template definition via ordinary HTTPS. No sign-in, installation or GitHub connection. The upstream open-library attribution was not independently checked; listing date is not publication date.
Define a coordinator and one owner per stage, record structured handoffs, batch human approvals and periodically review shared-computer and routine workloads.
What was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. Referenced engineer demos are author reports.
Review bot roles and skills, keep standing rules in a private handbook, and organize tasks around sources, owners, outputs and evidence before expanding the team.
What was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. Distinct from the audit prompt: this article also covers standing rules and acceptance.
A weekly pipeline separates scouting, analysis, strategy and building into narrow bot roles, using file handoffs and reusable instructions. Linked templates remain unchecked.
What was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. Linked templates and third-party tools remain unchecked.
Use webhook notifications for meaningful state changes so the responsible bot handles ready work, with explicit handoff, verification and human decision boundaries.
AppsGitHubWhat was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. Codex read the linked README: a playbook and helpers, not an operating service.
An architecture idea separates short facts, long GitHub documents and structured Supabase records, retrieving only what a task needs. No verified integration configuration is provided.
AppsGitHubSupabaseWhat was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. Supabase integration and usage benefits were not verified; architecture idea only.
An unofficial browser extension packages sourced page selections, images, tables and screenshots, with a material basket, send preview and sidebar follow-ups.
AppsChromeWhat was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. Codex read the repository README, including permissions; no installation or security audit.
Maintain a registry of service owners, paths, dependencies and recovery steps, with ordered scripts and human-triggered acceptance checks. The author says the full restore remains untested.
AppsGiteeWhat was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. The reset claim differs from official snapshot guidance. Only registry and acceptance ideas are recommended; private backups were not accessed.
The author describes a WeChat bridge that sends message IDs to a webhook, lets the bot read local inbox records and returns results, with deduplication and token-expiry handling.
AppsWeChatWhat was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. The protocol reference is not the tutorial original; bridge code and authorization were not checked.
Track agent tasks and artifacts by ID, then wake the bot with a webhook to inspect outputs and acknowledge the report. Persisted state supports recovery from missed notifications.
AppsPi AgentWhat was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. Codex also read the linked repository README, not implementation code.
A prompt audits existing roles, skills and routines against workflow needs and marketplace options, returning optimization or merge suggestions before any expansion.
AppsGrok Bot MarketplaceWhat was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. Marketplace listings and template installation were not checked.
A creator-tracking case separates bot conversations and data gathering from Codex CLI implementation and durable storage after an initial markdown-only result fell short.
AppsCodex CLIXWhat was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. No repository was provided; implementation outcomes are author reports.
Five author workflows use a bot email address for scheduling, newsletters, trial tracking, supplier quotes and file handoffs, with explicit fields and human approval boundaries.
AppsGrok Bot EmailWhat was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. All five author replies were obtained, improving on the October 15 hold; no mailbox access.
The author proposes a Notion MCP knowledge base for documents, prompts, skills and bot profiles across devices, reducing the need for local file control.
AppsNotionWhat was checked
Tie Zhu read the X source; Codex reviewed the handoff and deduplication, without independently re-reading X. Media and workflows were not tested. Notion configuration, permissions and synchronization were not verified.
Official troubleshooting covers sign-in, cloud-computer access, app authentication, attachments and routines, distinguishing retry, recovery, update and reset steps.
What was checked
Read the official page. October 12 is an update date, not publication. No permissions changed, external tests run, or computer updated/reset; guidance was not executed on the account.
Use a resource inventory, connectivity checks and recovery plan rather than treating one author's machine as a plan guarantee. Untested.
AppsGrok BotLinuxTailscaleWhat was checked
The collector read the body and viewed 12 images; 0 videos were not watched. The collector reports viewing 12 images; 0 videos remain unwatched. Images report the author's system, CPU/cgroup, memory, storage, GPU probes and permissions/tools. Missing GPU utilities alone do not prove GPU absence. Independent site exercises; no live workflow tested. Linked references were not all reread.
Review Jev triage in shadow mode. Public-source follow-up did not obtain the original script, complete eight cases or labels. Metrics remain one author demonstration, not a reproduced site result.
AppsGrok BotJevXWhat was checked
Retains prior body, seven images and 39.60s sampled video without audio. Second follow-up checks four identifiers, public code, limited recent author posts and associated directory pages, without attributable originals. Article-endpoint 404 is a failed reread, not evidence of deletion. Script, eight cases/labels, schema, thresholds and shadow code unavailable; no substitute or execution.
Define the user path, input, success and failure behavior, then tie tests, logs and screenshots to a code revision. Untested.
AppsGrok BotCursorGitHubWhat was checked
The collector read the body and viewed 1 images; 0 videos were not watched. The collector reports viewing 1 images; 0 videos remain unwatched. The image is a cover. The Lauren article and plugin page were not fully reread; this guide uses the acceptance framework from the read source. Independent site exercises; no live workflow tested. Linked references were not all reread.
24-source and six-video map, with archive classification, multiple-Webhook checks and further source tracing. Original Jev and Feishu role/bridge files remain missing; audio uncertain. Untested.
AppsGrok BotWhat was checked
Retains 24 bodies, 245 images and six sampled videos. Second follow-up checks the associated Feishu author account, 14 path trees and keyword results, not full code reviews; three monitor archives, publisher changelog/issues and an author multiple-Webhook reply. Jev originals, Feishu role/bridge source and audio remain unresolved. Failed article rereads do not establish deletion. No live reproduction.
Separate private-network reachability from SSH authorization and validate the host, user and revocation path. Untested.
AppsGrok BotTailscaleSSHWhat was checked
The collector read the body and viewed 5 images; 0 videos were not watched. The collector reports viewing 5 images; 0 videos remain unwatched. Images show two devices, an additional browser authentication step and the author's box user. Host, user and authentication rules remain environment-specific. Independent site exercises; no live workflow tested. Linked references were not all reread.
Connect source records, topic approval, drafting and fact review through a single low-risk sample. Untested.
AppsGrok BotXMarkdownWhat was checked
The collector read the body and viewed 19 images; 0 videos were not watched. The collector reports viewing 19 images; 0 videos remain unwatched. Images assign 01-inbox, 02-briefs, 03-drafts, 04-review and 06-analysis to separate roles. Writing uses supplied evidence and retrospectives use authorized data. The local-execution screenshot differs from the old text. Independent site exercises; no live workflow tested. Linked references were not all reread.
Inspect author Skill, routine and archives: separate imports, baselines, scans, failures and scheduled versus actual times. First-sighting formulas still differ. No standalone collector or delivery test.
AppsGrok BotXGitHubWhat was checked
Retains pinned 8ad0a75 Skill/routine and formula/failure checks. Additional 1236 import, 0604 baseline and 0907 failure records distinguish existing counters from new deliveries and schedules from actual times. File contents verified, underlying X data and delivery not reproduced. Git archiving is not counted as new. Original installation wording and standalone collector remain missing.
Separate role definitions, accessible memory, work files and recovery instructions, then test a small restore. Untested.
AppsGrok BotMarkdownGitWhat was checked
The collector read the body and viewed 13 images; 0 videos were not watched. The collector reports viewing 13 images; 0 videos remain unwatched. Images add registry.md, MEMORY.md, agent-spec.md and RESTORE.md, beginning with a user-approved roster. The 73KB, seven-role and 80% figures are author reports, not reproduced results. Independent site exercises; no live workflow tested. Linked references were not all reread.
Twelve author-reported usage tips on routines, context, CLI handoffs and retries. All ten images read; billing needs account-specific updates. External API and cloud-runtime claims remain untested.
AppsGrok BuildCursorXOpenRouterModelScopeWhat was checked
The collector read the body and viewed 10 images; 0 videos were not watched. The collector reports viewing 10 images; 0 videos remain unwatched. Images illustrate context management, CLI division and webhooks. Billing advice is account-specific and savings have not been reproduced. Independent site exercises; no live workflow tested. Linked references were not all reread. Legacy Cursor in-progress cap behavior was checked separately; it is not generalized to other accounts.
Share or paste links to a knowledge inbox with deduplication and completion checks. Adds author-reported multiple Webhooks and entry/field mapping suggestions. Full Shortcut, invocation and storage unverified.
AppsGrok BotShortcutsWebhookWhat was checked
Retains body, 12 images and two sampled phone videos. Collector read author reply 2111499978728448348 describing multiple instructed Webhooks for one Bot. Site-written routing, input-field and receipt checks are suggestions based on that statement, not a tested configuration. Full Shortcut and storage receipt unavailable; invocation uncertain. No SDK or extension substituted.
Check role setup, identity, execution environment and triggers separately before expanding a first workflow. Untested.
AppsGrok BotCLIWhat was checked
The collector read the body and viewed 26 images; 0 videos were not watched. The collector reports viewing 26 images; 0 videos remain unwatched. Screenshots add main-Bot settings, connectors, handoff receipts and local versus cloud CLI use. An unreadable small-text setting is not made into a universal instruction. Independent site exercises; no live workflow tested. Linked references were not all reread.
Inventory coordination, judgment, research, growth, production and monitoring, then review overlaps and missing artifacts. Untested.
AppsGrok BotMarkdownWhat was checked
The collector read the body and viewed 16 images; 0 videos were not watched. The collector reports viewing 16 images; 0 videos remain unwatched. Images identify sender, recipient, artifact and reviewer, keeping topic, author input, final copy and publication with the owner. Public installation packages were not run. Independent site exercises; no live workflow tested. Linked references were not all reread.
Validate a first Bot with a synthetic operating sheet, check its numbers, repeat with another input, then draft a skill and routine. Untested.
AppsGrok BotCSVWhat was checked
The collector read the body and viewed 7 images; 0 videos were not watched. The collector reports viewing 7 images; 0 videos remain unwatched. Images show handoffs, a skill menu and a routine; historical configuration screens are not treated as current menus. Independent site exercises; no live workflow tested. Linked references were not all reread.
Check supported environments and account types, distinguishing subscription login from API authentication and the index's mismatched description. Untested.
AppsClaude CodeGrok BotWhat was checked
The collector read the body and viewed 11 images; 0 videos were not watched. Index/source mismatch: the local page uses official access checks, not the original registration technique. Independent site exercises; no live workflow tested. Linked references were not all reread.
Translate recruiting, sales, marketing, procurement and design requests into evidence, deliverables and review points. Untested.
AppsGrok BotGmailFigmaWhat was checked
The collector read the body and viewed 1 images; 0 videos were not watched. The collector reports viewing 1 images; 0 videos remain unwatched. The only image is a cover and adds no operational steps. Independent site exercises; no live workflow tested. Linked references were not all reread.
Review a PR ledger against the current commit, checks, conflicts and comments before proposing follow-up. Untested.
AppsGrok BotGitHubNotionWhat was checked
The collector read the body and viewed 8 images; 0 videos were not watched. The collector reports viewing 8 images; 0 videos remain unwatched. Screenshots add ownership and Working/Watching/Ready-for-review states, distinguishing Agent completion from human merging. This guide keeps merging at a separate review point. Independent site exercises; no live workflow tested. Linked references were not all reread.
Compare access, existing subscriptions, usage pools and additional charges before deciding whether to switch. Untested.
AppsGrok BotCursorWhat was checked
The collector read the body and viewed 2 images; 0 videos were not watched. The collector reports viewing 2 images; 0 videos remain unwatched. The screenshot lists Free, Plus, Premium and Super, placing Bot under Premium. It does not establish Ultra or promotions; prices and discounts remain unverified. Independent site exercises; no live workflow tested. Linked references were not all reread.
Organize ideas with an editorial profile, source timestamps, visible metrics and original angles without inventing reach or copying posts. Untested.
AppsGrok BotXNotionWhat was checked
The collector read the body and viewed 1 images; 0 videos were not watched. The collector reports viewing 1 images; 0 videos remain unwatched. The only image is a cover and adds no operational steps. Independent site exercises; no live workflow tested. Linked references were not all reread.
Coordinate research and A/B drafts. Associated GitHub account and 14 path trees checked; full SOUL and bridge files remain unavailable, audio uncertain. Existing text/visual workflow untested.
AppsGrok BotFeishuXWhat was checked
Retains body, 22 images, 13.84s sampled video and automatic audio review. Associated liyue-aigc account, 14 complete default-branch path trees and SOUL/feishu/x-intel/lark content queries checked; two lark hits reference video Skill documents, not a bridge. Not a full code review of all repositories. Role/bridge files still missing; audio uncertain, no human listening or audio-derived steps. Untested.
Check key points, action items and evidence in authorized meeting notes before connecting tools or scheduling follow-up. Untested.
AppsGrok BotMarkdownWhat was checked
The collector read the body and viewed 12 images; 0 videos were not watched. The collector reports viewing 12 images; 0 videos remain unwatched. Images show approvals, conference photos and an anecdotal usage record. Historical eligibility screenshots are not current plan rules. Independent site exercises; no live workflow tested. Linked references were not all reread.
Use one intake and a ledger of owners, states and artifacts; propose a missing role before expanding the roster. Untested.
AppsGrok BotFeishuWhat was checked
The collector read the body and viewed 8 images; 0 videos were not watched. The collector reports viewing 8 images; 0 videos remain unwatched. Images show multiple roles and a Feishu competitor brief, illustrating consolidated outputs without establishing current prices. Independent site exercises; no live workflow tested. Linked references were not all reread.
Plan outputs and handoffs, then inspect the author publisher source: Markdown drafts, image paths, browser profile and field-level receipts. ok does not prove complete images. Static review; untested.
AppsGrok BotWhat was checked
Retains body, 11 images, 59s sampled video and pinned 6cd2f1b source scope. Changelog and empty public issues checked; generic image fixes do not document repair of the two recording failures. ok may still coexist with imgFail. Recording version, repair and current X compatibility unverified; no execution.
Clarify platforms, collection rules and output locations, then validate a sample across research, selection, writing and review. Untested.
AppsGrok BotFeishuWhat was checked
The collector read the body and viewed 18 images; 0 videos were not watched. The collector reports viewing 18 images; 0 videos remain unwatched. Images confirm source-to-topic-to-draft handoffs and human review. Two interview prompts were not obtained from the images; this guide writes its own questions. Independent site exercises; no live workflow tested. Linked references were not all reread.
An author walkthrough contrasts packaged plugins, remote MCP and stdio servers using Composio and sequential-thinking, distinguishing Grok Bot from chat and Build. Older UI paths need rechecking.
AppsComposioGmailsequential-thinkingPractice with a complete tutorialWhat was checked
Read the vendor's September 3, 2026 guide as historical backfill. No servers or accounts were connected. Current compatibility is untested; older menu paths and credential-handling examples are not adopted as our instructions.
A community skill uses Codex computer control to select Bots, inspect conversations and continue tasks, with separate rules for messages, files and routines. It requires a signed-in Grok Bot app on the same Mac.
AppsCodexWhat was checked
Read the README and full skill, including prerequisites and action boundaries. Not installed or tested. Repository timestamps are not treated as the document's publication date; this is historical backfill.
A role-profile collection covers productivity, marketing, sales, operations, personal tasks and development, with starter teams and OpenClaw/Hermes conversion links that exclude memory and local files.
AppsGmailSlackGitHubOpenClawHermesWhat was checked
Read the README and sampled the Inbox Triage profile. No team import or conversion was run. This is not a tested ranking; older plugin-menu paths need checking against current official guidance. Historical backfill.
A user keeps project-group status and tasks in Google Docs with a central index and manager Bot, while reporting coordination difficulties and high usage. A concrete setup to compare, not a proven efficiency gain.
AppsGoogle DocsPractice with a complete tutorialWhat was checked
Read the original post and visible discussion. Setup and difficulties are user reports; outcomes and bills were not verified. No release date was inferred from relative time, and separate Bots are not treated as permission isolation.
If Grok Bot won't operate your own computer, go to the computer section in Settings and set local execution to ask every time.
What was checked
Read the full post text via the X read API; author results not reproduced. The settings path is as the author describes; screenshots not viewed.
Describes Matt Palmer's Media Manager bot: after a YouTube draft and a Figma link, it transcribes via AssemblyAI, writes titles and timestamped descriptions, edits a Figma thumbnail, and writes back via the YouTube API.
AppsYouTubeFigmaAssemblyAIWhat was checked
Read the full post text via the X read API; author results not reproduced. Chinese secondhand write-up; original video and template not viewed.
Five-step guide: four Grok Bots (tweet collector, profile analyst, designer, engineer) share a project folder in a group chat to turn your recent tweets into a personal profile site, previewed before deploy.
AppsXVercelCloudflare PagesWhat was checked
Grok Bot read the article text and prompts; ~13 screenshots/recordings and linked posts were not viewed. Deployment is described, not demonstrated. Results not reproduced; summary, attribution and source link only.
The author has Grok Bot hand coding to a Cursor Cloud Agent on a GitHub repo that researches, codes and opens a PR with a preview link; nothing merges until the author approves.
AppsCursorGitHubPractice with a complete tutorialWhat was checked
Read the full post text via the X read API; author results not reproduced.
The author's pipeline: Slack bug reports become GitHub issues, go to a Cursor cloud agent, and PRs return to Grok Bot for human approval, with plans to let a chief-of-staff bot approve later.
AppsSlackGitHubCursorPractice with a complete tutorialWhat was checked
Read the full post text via the X read API; author results not reproduced.
A step-by-step article to hand the bot: install Orca on the Grok Bot computer as a remote server, connect Mac and phone through Tailscale to drive Claude Code there, with the human sign-in steps listed.
AppsClaude CodeTailscaleOrcaWhat was checked
The collector read the body and viewed 8 images; 0 videos were not watched. The collector reports viewing 8 images; 0 videos remain unwatched. The architecture connects an Orca service and CLI over Tailscale with separate desktop and phone pairing. Watchdog reliability has not been reproduced. Independent site exercises; no live workflow tested. Linked references were not all reread.
A hosting founder lists routines handed to Grok Bot: growing site listings, a daily malware hunt for customers, and keeping a site fresh; he calls routines cron for real work.
Practice with a complete tutorialWhat was checked
Read the full post text via the X read API; author results not reproduced.
Fifteen copy-ready use cases (competitor brief, invoice review, bug repro, PR review, Slack escalation) and a five-step loop: one bot, a precise job, the prompt, one plugin, a reversible run before scheduling.
AppsSlackGitHubNotionWhat was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced.
Ben Lang's roundup of team tips: bots as roles, channels per project, routines not too frequent, give recurring work to a fresh bot, approval rules for writes, past examples, and a Chief of Agents for shared rules.
What was checked
Read the full post text via the X read API; author results not reproduced. A secondhand roundup of team tips; the author is on the SpaceXAI team.
The author's steps: install the app, sign in with a Cursor account, have the bot set up an SSH server plus Cloudflare Tunnel or Tailscale on its VM, then reach it remotely.
AppsTailscaleCloudflareWhat was checked
Read the full post text via the X read API; author results not reproduced. Plan details reflect launch week.
Trial steps from launch week: download the client, sign in, pick One Week Trial on the Get Started page, activate, and add a card when prompted.
What was checked
Read the full post text via the X read API; author results not reproduced. Flow as of 2026-08-12 and may have changed; images not viewed.
A 10-step setup map: official install and data settings, one job, approval rules, minimum access with read-only checks, output format, first task, verification, preferences, and a reusable skill.
What was checked
Read the full post text via the X read API; author results not reproduced. It is an outline; detail per step is limited.
After reading the official founder-led sales guide (a chief of staff plus 8 bots), the author condensed it into one FounderSales bot that researches prospects, drafts outreach and tracks follow-ups, with approval.
What was checked
Read the full post text via the X read API; author results not reproduced. Template is in replies and was not opened.
The author built a bot that forwards a question to Google Gemini and returns a labeled answer, using free Gemini for simple questions; the API key goes in via a secure secret card, never the chat.
AppsGeminiWhat was checked
Read the full post text via the X read API; author results not reproduced. Video and template in replies not viewed.
A bot's title is display text next to its name; it doesn't sort or route, but works as a role label. Set it in settings or Edit Profile, and use sidebar sections, including movable Unassigned, to group bots.
What was checked
Read the full post text via the X read API; author results not reproduced. Video not watched.
As of version 0.68.1 the webhook trigger detail sometimes won't open to show the URL and secret; the author's workaround is asking Grok Bot for a link in chat that opens the detail panel.
What was checked
Read the full post text via the X read API; author results not reproduced. Version-specific issue that may since be fixed.
The author installed Grok Build inside the bot and sends heavy code edits, refactors and long-context analysis to the terminal tool, leaving connectors and posting to the bot to save tokens.
AppsGrok BuildWhat was checked
Read the full post text via the X read API; author results not reproduced. Savings are self-reported.
A long Chinese primer: bots, the cloud computer and routines, plans and usage, templates/plugins/skills, nine use types, settings for users in China, and 30 rules from the official docs, with sources.
What was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced. Plan and price details are as of 2026-10-11 per the author and may have changed.
Ten high-star SKILL.md repos (Superpowers, Anthropic skills and more), with the tip to paste a repo URL and ask the bot to add it, and a warning to read skills first since they run with bot access.
AppsGitHubWhat was checked
Read the full post text via the X read API; author results not reproduced. Repos not opened; star counts are the author's.
A Gmail triage flow: connect Gmail, label mail as Action, Waiting, Money, Receipts, Newsletters or Keep, save replies to Drafts, and the user presses send.
AppsGmailPractice with a complete tutorialWhat was checked
Read the full post text via the X read API; author results not reproduced. Attached image not viewed.
Hiding a bot only removes it from the sidebar while its routines keep running; deleting removes routines but files and sign-ins stay on the shared computer, so retire bots in order.
What was checked
Read the full post text via the X read API; author results not reproduced. The four-step card image was not viewed.
Meetup notes: a coordination layer keeps specs and acceptance criteria in the repo; Grok Bot delegates code, PRs and evidence to Cursor or cloud agents. SDD has three tiers; humans keep production calls.
AppsCursorGitHubPractice with a complete tutorialWhat was checked
Read the author's full self-thread in the same conversation; author results not reproduced. These are attendee notes; the original speaker is not named; images not viewed.
A team shares one Grok Bot via Telegram: an in-app setup bot owns the bridge script, tokens and allowlists; a worker bot turns order notes into PDFs in the thread and refuses config commands.
AppsTelegramWhat was checked
Read the author's full self-thread in the same conversation; author results not reproduced. The author describes it as an unofficial setup of unknown durability.
A remodeling bot template: purchase, receipt and subcontractor trackers, receipt-wins money rules, draft-only outbound mail, Gmail folder sync, and Mon/Wed/Fri status reports. One bot, one job.
AppsGmailPractice with a complete tutorialWhat was checked
Read the full post text via the X read API; author results not reproduced. Template link not opened.
The author's next-step setup: install the last30days skill for research and waza for dev and writing standards, give agents a dedicated browser, and run pi agent with other models to save usage.
What was checked
Read the full post text via the X read API; author results not reproduced. Part one was not read; all are third-party skills.
The author's four work bots: a read-only Inbox Manager that triages email, Slack and DMs, a Calendar EA that protects focus time, plus information and execution bots, each with a template link.
AppsGmailSlackGoogle CalendarPractice with a complete tutorialWhat was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced. Template links not opened.
The author's article covers limits of the built-in @mail.grokbot.com inbox (no rename, shared reputation, no inbox view) and connecting an own-domain mailbox via MCP, webhook triggers, and draft-only first.
AppsShipmailWhat was checked
Read the post and the linked blog in full; the author founded Shipmail and the approach uses his paid product, so it is commercially interested; not reproduced.
The author has Codex drive the Grok Bot window to review roles, chats, routines, plugins and handoffs, finding gaps, duplicate routines and safety issues, asking before publishing or deleting; prompt included.
AppsCodexWhat was checked
Read the full post text via the X read API; author results not reproduced.
Five beginner mistakes and fixes: treating it as a chatbot, over-broad access, building a team on day one, no brakes. Prefer connectors, @-tag bots, and note hidden bots keep running routines.
What was checked
Read the full post text via the X read API; author results not reproduced. Part of the final item was cut off; setting names are as the author describes.
A checklist route: write what the company does and must never do alone, connect mail and calendar, start one draft-only Chief of Staff, save a good run as a skill, add routines later, then split into bots.
AppsGmailGoogle CalendarWhat was checked
Read the full post text via the X read API; author results not reproduced. The quoted post was not read.
The author's guide to connecting a set of Grok Bots to Feishu (Lark), so tasks can be assigned and results received in Feishu, with the prompt and configuration steps to give the bot.
AppsFeishuWhat was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced. Unofficial integration; images not viewed.
The author connects Grok Bot to WeChat in three steps: send a prepared prompt to the bot, set up a WeChat channel with the needed values, and scan a QR code to bind, then direct bots from WeChat.
AppsWeChatWhat was checked
The collector read the body and viewed 12 images; 0 videos were not watched. The collector reports viewing 12 images; 0 videos remain unwatched. Images show a routine, a separate relay and secure credential input, with Notion ideas awaiting review rather than X publication. Bridge compatibility and official support remain unverified. Independent site exercises; no live workflow tested. Linked references were not all reread.
A monthly routine audit: list fired routines, count approved outputs, flag silent failures or ignored ones, nominate one to remove, and wait for human approval before deleting.
What was checked
Read the full post text via the X read API; author results not reproduced.
Usage tips from launch-event day 1: browser actions cost most, prefer connectors, audit routines, map roles for 1-2 hours first; an hourly expense routine was moved to Fridays at 8am.
What was checked
Read the full post text via the X read API; author results not reproduced. Secondhand from the event; livestream not watched.
How to use skills: after a process works once, ask the bot to save it as a skill in generic steps without hard-coded links or accounts, then simply ask it to run that skill.
What was checked
Read the full post text via the X read API; author results not reproduced.
A SpaceXAI engineer on how the team builds with Grok Bot: managing coding agents like a capable intern, keeping work moving while away, learning how you work, with concrete setup and flows.
AppsCursorGitHubSlackPractice with a complete tutorialWhat was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced. The author is a SpaceXAI employee; personal post, not official docs.
Ten copy-ready role prompts (SDR, support, recruiter, analyst, researcher, chief of staff) that all stop at a draft, plus five safety lines: never send or pay, ask first, dedicated accounts, no keys.
What was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced. Most prompts were read as code blocks; some content is in images not viewed.
Grok Bot's inbox supports plus aliases (+newsletters, +receipts); routines can listen to one alias each to summarize or file bills, and a spammed alias shows who leaked the address.
What was checked
Read the full post text via the X read API; author results not reproduced. The quoted post was not read.
Using the Usage page as a baseline, the author stops or slows routines nobody reads, removes extra bot hops, and trims memory, skills and stale rules, packaged as an installable Sweeper template.
What was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced. The author is a SpaceXAI Ambassador; the template was not installed.
The author cut 15-minute inbox, meeting and ticket routines to one or two passes a day because frequent runs added noise and token cost; when present, call the bot manually.
What was checked
Read the full post text via the X read API; author results not reproduced. Two shorter posts by the same author were not added separately.
Matt Palmer's five-step start: connect Slack, Gmail and Notion for retrieval, install mobile, build workflows like sheet updates and unsubscribes, then turn daily work into routines and triggers.
AppsSlackGmailNotionFigmaWhat was checked
Read the full post text via the X read API; author results not reproduced. Author is a SpaceXAI employee; personal post, not official docs.
On a plugin's page, click the link icon to get a grokbot:// install deeplink anyone can tap; the author notes these links only work on X desktop clients.
What was checked
Read the full post text via the X read API; author results not reproduced. The author is a SpaceXAI employee; personal post.
A five-part bot setup framework: purpose, plugins, output destination, routine cadence, and a manual test before automating; example saves LinkedIn leads to Google Sheets each morning.
AppsLinkedInGoogle SheetsWhat was checked
Read the full post text via the X read API; author results not reproduced.
A language-learning bot template: pick a language and level for a morning lesson, a midday voice drill, an evening quiz plus a longer Sunday one, an 8-week plan with tracking, and practice anytime.
What was checked
Read the full post text via the X read API; author results not reproduced. Template link not opened.
How to customize voice: open Voice Mode from the voice icon, adjust speaking speed (up to 2x) and language in Settings, choose a voice and personality, and leave language detection on Auto.
Practice with a complete tutorialWhat was checked
Read the full post text via the X read API; author results not reproduced. Image not viewed; setting names are as the author describes.
Management skills for running several bots: set priorities before delegating, define outcomes and limits, match work to ability, add checkpoints, distrust fluent answers, review and fix the process.
What was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced.
A paste-in prompt: review 90 days of email, calendar and notes, find missed customer follow-ups, exclude active or declined ones, rank them, draft messages for review without sending, then make it monthly.
AppsGmailGoogle CalendarWhat was checked
Read the full post text via the X read API; author results not reproduced. Image not viewed.
Run a job manually, fix the process, save it as a skill, test new inputs, define failure handling (unavailable source, stale data, partial runs, approvals), then test the routine before trusting it.
What was checked
Read the full post text via the X read API; author results not reproduced.
Four workflows summarized from Grok Bot 101: an X-follows Notion CRM, a fitness coach bot, gathering Slack/Notion/GitHub context for a coding agent, and cross-app internal search.
AppsNotionSlackGitHubCursorPractice with a complete tutorialWhat was checked
Read the full post text via the X read API; author results not reproduced. Secondhand summary of the official guide; the guide itself was not checked here.
A bot is essentially name, title and description; start from a voice note and describe the workflow concretely. Example: Tech Demos scans X bookmarks each weekday, drafts a prompt, and starts a cloud agent once approved.
AppsCursorWhat was checked
Read the full post text via the X read API; author results not reproduced.
Three ways to interact with a bot: chat, routines and triggers (schedules or Slack and GitHub events), and other bots messaging and triggering each other.
AppsSlackGitHubWhat was checked
Read the full post text via the X read API; author results not reproduced.
An article on turning the Grok Bot cloud computer into an always-on private server with a few commands and one prompt, reachable over SSH with a VNC desktop.
What was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced. Community approach; may need redoing when the computer is reassigned.
A coordinator template for bot teams: probes connectors before relying on them, hands work to a Lead only after readback, reviews even partial deliveries, and reports in Markdown; full skill text is in a repo.
What was checked
Read the full post text via the X read API; author results not reproduced. Template and repo not opened.
The author's 21 hacks: plan first, layered instructions, screen-recorded teaching, separate inboxes, phone calls, cookie sync, CLIs for frequent services, and pasting the article into your bot to set it up.
AppsAgentMailTwilioTailscaleWhat was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced. Several tools mentioned are the author's own projects.
Lauren Tan of SpaceXAI on using bots for the first and last mile: connect calendar, CRM, Slack and email to give context instead of micromanaging, and use scheduled or event routines to follow through.
AppsSlackPractice with a complete tutorialWhat was checked
Read the full post text via the X read API; author results not reproduced. The latter part of the long post was truncated in the API response.
Subscribe routines to Datadog, Sentry or Slack events so a backend alert spins up bots to investigate and triage, rather than relying only on daily schedules.
AppsDatadogSentrySlackWhat was checked
Read the full post text via the X read API; author results not reproduced. A near-identical second post was treated as a duplicate.
A job-hunt bot template: five matched roles each weekday with quotes, tracker updates from Gmail, tailored resumes, interview prep from Calendar, and a Sunday review; it never applies or sends.
AppsGmailGoogle CalendarWhat was checked
Read the full post text via the X read API; author results not reproduced. The template itself was not installed or inspected.
A detailed guide to SSH from macOS, Windows or Linux into the Grok Bot Linux computer over Tailscale, using scp, logs and CLIs; noted as unofficial and possibly needing reinstall after updates.
AppsTailscaleWhat was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced.
Add a LinkedIn connector such as Composio or LinkMCP via plugins, authorize with OAuth rather than a password, test a low-risk action first; direct browser login risks bans and sends need approval.
AppsLinkedInComposioPractice with a complete tutorialWhat was checked
Read the full post text via the X read API; author results not reproduced. Not tested; plugin names are the author's.
With no native Apple Calendar action, the author bridges it: the bot stores the event and sends an email, a small Cloudflare Pages bridge holds it, and an iPhone Shortcut pulls it into Calendar.
AppsApple CalendarCloudflareAgentMailWhat was checked
Read the X Article body via the X read API; images not viewed; author results not reproduced. Prompt and Shortcut code are images and were not viewed.
Steps to let the bot operate a Windows PC: check an eligible plan, install the official desktop app, enable local computer access in settings, with a warning that file access and local commands are high risk.
AppsWindowsWhat was checked
Read the full post text via the X read API; author results not reproduced. Plan and setting names are as described on 2026-08-22 and may have changed.
Suggests beginners start with official engineering templates in the Marketplace: add one close to your work, inspect bundled skills, routines and connectors, then adapt it; coding can go to Cursor.
AppsCursorGitHubWhat was checked
Read the full post text via the X read API; author results not reproduced.
Summarizes Lingxi Li's setup: five domain bots turn tasks into Cursor cloud agents with expected proof, a manager bot runs daily 1:1s and postmortems, timed PR checks, and low-risk PRs auto-merge.
AppsCursorGitHubPractice with a complete tutorialWhat was checked
Read the full post text via the X read API; author results not reproduced. Third-party summary of a workshop; video not watched.
A skill packaging Lauren Tan's talks: read code first, reproduce bugs, run real user flows, use feature maps to verify, and turn repeated review comments into types, lint or CI.
What was checked
Read the full post text via the X read API; author results not reproduced. Skill file not opened; third-party packaging.
A Chinese beginner guide: install and sign in, which plans work and sign-in pitfalls, the author's bot framework (infrastructure, model choice, advisor, bot and org tuning) with template links and naming rules.
What was checked
Read the available post body; part of the long post was unavailable. Summary covers the sections read. Prices are dated 2026-09-10; templates not opened; not reproduced.
Three start steps: define the job, run once manually, then add a routine; require approval for posting, paying and deleting; inspect a template's skills, routines and plugins before and after installing.
What was checked
Read the full post text via the X read API; author results not reproduced.
The author had Grok Bot read every pstack SKILL.md, check the Grok Build skills docs, swap out Cursor-only parts, and produce 46 installable skills in about 25 minutes.
AppsGrok BuildWhat was checked
Read the full post text via the X read API; author results not reproduced.
The author's split: keys, builds and private files stay local; the Grok cloud computer handles 24/7 page watching, schedules and tests; a personal VPS keeps sites and long-running services.
What was checked
Read the full post text via the X read API; author results not reproduced.
Two no-code templates: Site Sentry, a read-only site and deploy monitor that redacts secrets, and an AI Security Advisor that reviews prompt injection and tool abuse. Lesson: write the job like a hire.
Practice with a complete tutorialWhat was checked
Read the author's full self-thread in the same conversation; author results not reproduced.
A seven-Bot content team covering social repurposing, a chief of staff, invoice hunting, topic research and weekly SEO reports, with the tools each one uses.
AppsGmailGoogle CalendarTodoistNotionTypefullyGoogle DrivePractice with a complete tutorialWhat was checked
Read the author's article and workflow description; the workflow was not reproduced.
A preprint describes a supervised Grok Bot research team working on stellar spectra. No matched human-team comparison was performed, so its timeline does not establish a productivity gain.
What was checked
Checked the arXiv abstract, authors and submission date; the study was not reproduced or treated as peer-reviewed.
Building an inbox manager with the Gmail plugin and browser automation: labels, filters, a 7 am brief in Slack, and why to stay drafts-only for the first week.
AppsGmailGoogle CalendarSlackPractice with a complete tutorialWhat was checked
Read the author's article and workflow description; the workflow was not reproduced.
A roundup of ten plugins: Gmail, Slack, Notion, GitHub, Google Calendar, Firecrawl, HubSpot, Canva, Sentry and Composio’s own. Published by Composio.
AppsComposioGmailSlackNotionGitHubGoogle CalendarFirecrawlHubSpotCanvaSentryPractice with a complete tutorialWhat was checked
A roundup published by a plugin vendor; read in full, plugins not installed or tested.
A vendor guide connects Google Sheets through Composio and outlines worksheet creation, range reads, batch updates and keyed row updates. Available tools and permissions require account-specific checks.
AppsGoogle SheetsComposioPractice with a complete tutorialWhat was checked
Read setup and spreadsheet tool descriptions; no account was connected and no sheet was changed. No reliable release date was shown. Older plan lists, tool counts and timing claims are not adopted.
A Composio walkthrough connects and names separate Google accounts, checks account routing, and combines inbox or calendar tasks. It also covers Workspace restrictions and connection management.
AppsGmailComposioGoogle CalendarGoogle DriveGoogle DocsGoogle SheetsPractice with a complete tutorialWhat was checked
Read the author's guide and verification steps; no Google authorization or workflow test was performed. This October 7 guide is a category backfill, not today's news. Timing and native-account-limit claims are not adopted.
Composio's guide walks through adding its plugin, authorizing access and connecting Notion for page and task-database work. Check available actions and permissions in your own account.
AppsNotionComposioPractice with a complete tutorialWhat was checked
Read the vendor's setup and tool descriptions; no authorization or actions were performed. Its older plan list and speed claims are not adopted here.
The Help Center page on plugins such as Gmail, Google Calendar, Sheets, Drive and Slack, what each can and cannot do, and how to fix sign-in problems.
AppsGmailGoogle CalendarGoogle SheetsGoogle DriveSlackNotionZoomWhat was checked
Read the official page.
A Chinese tutorial builds a DataCamp learning Scout, checks course details and prerequisites, then saves the process as a skill and weekly routine. Includes corrections from the author's tests.
AppsDataCampWhat was checked
Read DataCamp's Chinese tutorial; credited the publisher because no individual byline was visible. August 30 is an update date, not an inferred release date. Not reproduced.
A long hands-on review from an indie developer, covering a daily blog check routine, batch updates through Cursor cloud agents, and lessons on usage and browser automation.
AppsCursorGitHubCloudflarePlausiblePractice with a complete tutorialWhat was checked
Read the author's article and workflow description; the workflow was not reproduced.
A Chinese guide bridges external A2A agents to Grok Bot webhooks and callbacks. It documents deployment, task state, authentication and retries, with companion connection instructions for a self-hosted server.
AppsA2A BridgeWhat was checked
Discovered via the author's LINUX DO post; read the repository README, Chinese guide and companion skill. The introduction is not listed separately. No installation, tests or security audit were performed; no release date was inferred.
Community role and team templates for code review, inbox triage and research, with inputs, deliverables and permission boundaries. Its readiness score is the project’s own checklist.
What was checked
Read the README and role catalog overview; templates were not imported and project scores are not our test results.
A community catalog for writing, design, SEO and Bot workflows, with examples, tool requirements and source notes. Check companion files and upstream licenses before using a skill.
What was checked
Read the README and catalog; installers were not executed and individual Grok Bot compatibility was not tested.
A writing skill in the Grok Bot catalog with editing and pattern-review workflows. It aims to retain the writer's voice and credits Peter Yang's original project.
What was checked
Read the skill file and repository credits. This links to a specific workflow rather than the catalog. Not installed; environment paths and companion dependencies require review.
A skill for reviewing Grok Bot routines for excessive frequency, duplicate work, long chats and empty-run noise. It proposes changes and requires environment-specific paths.
What was checked
Read the skill and its credit to Lauren Tan. No user Bot data was accessed and no audit was run; tool and path compatibility is unverified.
Architecture patterns for Bot collaboration: structured handoffs, shared-computer boundaries, stop conditions and human review. The repository is a reference with validators, not a hosted service.
What was checked
Read the project scope, limitations and examples; validators and deployments were not run.
A Chinese walkthrough combining Notion tables with focused Bots and reusable skills for orders, inventory, daily summaries and shifts. A small-team workflow reference, not a formal ERP.
AppsNotionWhat was checked
Read the author's walkthrough, skill examples and scope limits. We did not connect Notion or reproduce the workflow.
The author shares a research Bot that combines web, X, YouTube and GitHub sources into cited reports, with a public template link. Useful as a role-design reference.
What was checked
Read the original post and opened its template preview; the Bot was not imported or run, and research accuracy was not tested.
A first-person account of splitting review, coding, task tracking and coordination across a Bot group. We include the workflow idea, not the post’s unverified model-routing claims.
What was checked
Read the author’s post; screenshots were not independently authenticated and outcomes were not reproduced.
Eight daily routines from a creator, including competitor YouTube tracking, cross-platform follower counts, sales call reviews, prospect research and a Bot team standup.
AppsYouTubeNotionGoogle SheetsWhat was checked
Read the author's article and workflow description; the workflow was not reproduced.
A user describes a Bot team spanning coordination, personal assistance, writing, inbox triage and job search. This is a personal account, not official guidance or a guarantee of results.
Practice with a complete tutorialWhat was checked
Read the role and handoff descriptions; account permissions, bills and outcomes were not verified. Practices such as automated OTP handling were not adopted.
How a community operator set up a Bot team: turning off send permissions in connector toggles, a middle-manager structure, email and calendar routines, and limits such as uninspectable memory.
AppsGmailGoogle CalendarSlackKitWhat was checked
Read the author's article and workflow description; the workflow was not reproduced.
X explains how @bot forwards a post request to a connected Grok Bot and describes search, timeline and mention tools available through its official plugin.
Practice with a complete tutorialWhat was checked
Read the X Help Center. This is official documentation, not an indexed X post; no account was connected.
Official guidance on the shared cloud computer, browser sessions, connector plugins and durable files, including app setup and public versus signed-in X access.
AppsXPractice with a complete tutorialWhat was checked
Read the official guide, last updated October 12, 2026. No plugins were installed or account permissions inspected.
Official instructions for roles, Bot creation, duplication, template sharing and collaboration. Shared templates create copies without transferring account logins or chat history.
What was checked
Checked the official role, duplication and sharing sections; no third-party template was imported.
The official access expansion lists eligible SuperGrok and Cursor plans and explains separate Bot usage. Recheck current eligibility before subscribing.
What was checked
Read the official announcement dated August 26, 2026; the collection date is not its release date.
Official guidance compares desktop routing with Enterprise Team Setup for private access, using Tailscale and Cloudflare Tunnel examples. It states plan requirements and the need to pilot the Cloudflare pattern.
AppsTailscaleCloudflare TunnelWhat was checked
Read methods, prerequisites and limitations, last updated October 12, 2026. No clients or private networks were configured. The source says its Cloudflare example lacks the same team-computer testing as Tailscale.
Official guidance covers accounts, interface language, routine time zones, plugins, usage and notifications, distinguishing shared settings, individual Bot preferences and desktop-local execution controls.
What was checked
Read the official page, last updated October 12, 2026. No account, notification or permission settings were changed; the update date is not treated as a release date.
Official guidance on turning a successful task into a skill and a scheduled or event-triggered routine, including demonstrations, run history and failure handling.
Practice with a complete tutorialWhat was checked
Read the official guide, last updated October 12, 2026; workflows were not run in our account.
The official guide collection on how marketing, sales, support, recruiting, engineering, legal and PM teams run groups of Bots, with real workflows.
Practice with a complete tutorialWhat was checked
Read the official page.
The creator’s description covers chief-of-staff coordination, shared memory, ClickUp logs, demonstrations and routines, with chapters for navigating team-management tips.
What was checked
Checked the publicly indexed creator description and chapters only; the watch page was throttled. The full video and workflows were not tested.
The creator outlines Bot setup, YouTube research, X scouting, email triage and travel assistance. Use the chapters to choose a first task; results and costs need account-specific checks.
What was checked
Checked the publicly indexed creator description, chapters and publication date; the watch page was throttled and the full video was not watched.
Launch-day coverage with pricing for individuals and teams, comparisons with Claude and OpenAI agents, and open concerns such as no public benchmarks and opaque model routing.
Official security guide covering approval cards, Auto Review rules, Take over for passwords and codes, and the shared cloud computer across all Bots.
Official quick start covering eligible plans, desktop install on macOS/Windows/Linux, linking SuperGrok, and creating your first Bot and task.
Practice with a complete tutorialThe official announcement covering always-on cloud teammates, cross-app work, routines learned from a demonstration, and multi-Bot collaboration.
Browse by platform and use case. Each entry includes a note on what was checked. Video entries disclose when only the creator’s description was checked. Unverified items and possible reposts are held for review.