Team planner Bot: turn a one-line request into a reviewable roster

Define one team’s outcome, smallest roster and handoff contract before checking templates and applications.

The source demonstrates expanding a teaching template into three teams. This role card plans a roster while checking template availability and app capabilities separately.

Verification status

Source reading and media coverage are listed below. This site has not executed the live workflow. Fixtures, fields and prompts here are independent exercises, not the author’s results.

Scenario and input

Synthetic goal: one weekly content brief with no established collection workflow.

1. Narrow the outcome

Choose one near-term deliverable with an example of success, a deadline and a bounded scope.

2. Draft roles first

Give each role one main artifact, explicit inputs, stop conditions and a recipient.

3. Inspect teaching templates

Inspect what a template contains, including external apps and triggers, before treating it as runnable configuration.

4. Check apps by task

List necessary connectors, identities and read tests. A recommendation does not establish availability or authorization.

5. Expand after approval

Complete one manual brief before saving a template or adding specialists. Separate planning from actual creation and scheduling.

A synthetic acceptance fixture

roster.csv: role, input, deliverable, reviewer, trigger. Start with a researcher and human review rather than creating unrelated website and video teams.

Copyable task brief

Copy template

Use the synthetic fixture to prepare the deliverable for Team planner Bot: turn a one-line request into a reviewable roster. Do not connect real services yet.
Input: Synthetic goal: one weekly content brief with no established collection workflow.
Requirements:
1. Choose one near-term deliverable with an example of success, a deadline and a bounded scope.
2. Give each role one main artifact, explicit inputs, stop conditions and a recipient.
3. Inspect what a template contains, including external apps and triggers, before treating it as runnable configuration.
4. List necessary connectors, identities and read tests. A recommendation does not establish availability or authorization.
5. Complete one manual brief before saving a template or adding specialists. Separate planning from actual creation and scheduling.
Return itemized records, evidence locations, unknowns and next steps, marking unperformed actions explicitly.
Do not install tools, change access, purchase, publish, create Bots or schedule work in this pass.

Acceptance and review

The plan has one outcome and explicit artifacts and owners, without claiming uncreated roles are running.

Copy template

Review the artifacts without taking external actions.
Acceptance: The plan has one outcome and explicit artifacts and owners, without claiming uncreated roles are running.
Separate pass, fail, untested and evidence; distinguish synthetic fixtures from live results.
Stop on missing sources, permissions or uncertain outcomes and list the required evidence.
Propose one next step with a recovery plan and wait for confirmation.

Video supplement: give the publishing role a checkable X Article draft

The 59-second recording shows a publishing coordinator preparing an X Article from local Markdown and images. It is a role demonstration rather than team creation, and adds a workflow absent from the body and images.

Time	Visible evidence	Delivery meaning
00:00–00:14	The Bot reports using x-article-publisher; the user supplies a local Markdown path	Identify the article and assets; installation is reported in the recording
00:14–00:25	Local-command approval appears; copying assets encounters failures and retries	Command approval and asset completeness require separate checks
00:25–00:43	The user signs in to X in the Bot computer browser and returns control	The user handles the website session; installing a connector does not complete this login
00:43–00:58	The Bot returns a draft link, title, cover and an 8-of-10 image result	A draft exists in the demonstration, with two missing images; publication is not shown

Turn the role into five delivery steps

1. List the inputs. Supply an exact Markdown path, title, cover and expected image list, with draft creation as this run’s limit.
2. Check execution conditions. The recording uses a third-party skill and browser automation. The author repository is now available and key scripts were reviewed statically, as detailed below. Inspect commands, execution location, profile and affected files before any reproduction.
3. Check asset completeness. Report copy failures individually. Copying ten images does not establish that all ten reached the editor.
4. Let the user handle sign-in. Use the computer takeover card for login and verification, then return control, following the official takeover guidance. Account details and verification codes are omitted.
5. Review the draft receipt. Include the draft link, title, cover, expected and actual image counts, missing assets and unfinished work. An 8/10 result needs repairs and another preview before a separate publication decision.

Copy template

Write a Markdown-to-X-Article draft plan for a publishing coordinator; return the plan first.
Inputs I will supply: Markdown path, title, cover, image list and a reviewed tool source.
List missing inputs, proposed commands and execution location, and user takeover steps.
The draft receipt must include its link, title, cover, expected/actual image counts and missing assets.
Mark unavailable source code as unavailable; do not install a similarly named tool.
Mark an uncreated draft as not executed.
Do not run commands, sign in, upload assets or publish in this planning run.

Video verification scope

The collector checked approximately one frame per second across the timeline, including both ends, used 10fps frame differences to locate changes, and enlarged key frames. The largest viewed-frame gap was one second; this was not continuous playback. Some transitions and cropped message endings remain unreadable. Full-track automatic recognition reported music only; no human listening check was performed. Draft results are the author’s demonstration, not a site reproduction. The skill source is now available; repair of the two images in the recording remains unverified. The recording’s claim about X plugin limitations was not independently verified as a current general rule.

Source supplement: audit the publishing receipt against actual files

The author’s xaiwind/x-article-publisher is now available, pinned to 6cd2f1b. The account links to @xaiwind, and its system Chrome, playwright-core, skill directory and draft output match the recording. The commit actually used in the recording is unverified. Its LICENSE declares MIT and retains upstream attribution.

Apply five source-backed steps

1. Pin the tool and inputs. Record repository, commit, Markdown path, title and assets. The entry point is auto-publish.js, with dependencies in package.json. Verify actual Articles access against X’s official guidance.
2. Resolve image paths first. payload.js checks local paths and Obsidian attachments; this entry point rejects HTTP remote images. Eligible PNG/JPEG files may use macOS sips compression; other platforms skip it. Do not assume every asset was compressed.
3. Define the browser-session scope. The user signs in during a headed first run before considering --headless. --profile selects a separate session directory. The code may close an instance sharing that profile and clear lock files, so identify affected windows and execution scope first. This entry point has not been run here.
4. Inspect individual receipt fields. xpage.js counts imgOk, imgFail and atomicFail; failed images, code or tables may degrade to text. The function can still return ok: true, which does not establish complete images or formatting.
5. Reopen the draft for acceptance. The entry-point ending waits for autosave, cleans markers and returns a draft URL, without a separate server-content readback. Reopen it and check title, cover, image order/count, format degradation and leftover markers before a human publication decision.

A synthetic receipt with ok=true, summary.imgOk=8 and summary.imgFail=2 means a partial draft needing repairs. This explains the recording’s missing-image state; it does not prove this site repaired those two images.

Copy template

Review a publishing receipt without running the tool.
Source: xaiwind/x-article-publisher at 6cd2f1b752afe966c00bb803c1a1c5eb7a1085d4.
Ten images are expected; the synthetic receipt has ok=true, summary.imgOk=8 and summary.imgFail=2.
Check title, cover, image order/count, atomicFail, leftover markers and a reopened draft.
Separate confirmed facts, failures, readback needed and untested items.
Do not approve from ok or a URL alone. If no real draft exists, return a checklist.
Do not invent image repairs or publication results.

Source verification status

The skill description, dependencies, entry point, image handling and relevant page-write logic were reviewed statically. Source files and locations are linked rather than republished. No skill installation, browser launch, X draft write or current-interface reproduction occurred. Source availability is resolved; the recording’s exact version, its two image repairs and current X connector capabilities remain unverified.

Second source follow-up: a general fix is not this draft’s repair

The pinned CHANGELOG.md describes general image-order, missing-image and marker fixes. The public issues list was empty when checked. Neither source links a repair to the two failures in the recording or identifies its exact running version.

Keep both missing images on the receipt, then reopen the draft to check count, order and formatting after any repair. This is a source follow-up, without a repair receipt for that recorded draft.

Sources and applicability

Original article: @xaiwind; published 2026-09-16.

Index-to-source mapping: strong content match.

The collector reports viewing 11 images and reviewing 1 video(s) through sampled frames, without continuous playback. Screenshots show the author creating three groups and nine Bots; template and plugin availability still need checking.

Prices, interfaces and menus change. Check unresolved conditions against the actual account and current official documentation.

See the rewrite and verification map for all 24 sources