Grok Bot × Feishu: connect X research and drafts through a handoff table

Verify app, user and document permissions separately, then test one source-to-draft handoff through the message bridge.

The source organizes a coordinator, researcher and writer in Feishu. Its screenshot uses iMessage as a routine name; the actual function is a Feishu-group-to-webhook bridge.

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.

EXECUTION & TAKEOVER

Execution and takeover boundaries

The actions below require the agreed inputs, accounts and scope. Configuring a Routine does not expand that authorization.

Can run automatically

Organize supplied sources, research, drafts and handoffs with task ids, read scope, A/B versions and review state.

Requires confirmation

Confirm bridge configuration, app or user authorization, document sharing, role creation, test or production writes, group messages and publication separately.

Take over after a failure

Stop at the missing app, user or document permission layer for user takeover. Inspect acceptance and completion by task id, suppress echoes and avoid blind resends.

Scenario and input

Synthetic material: one read public X article and a designated test document.

1. Map the message path

Identify the group, app bot, relay, webhook routine and Grok Bot. Distinguish a group posting webhook from an app that subscribes to messages.

2. Check three permission layers

Check app scope, user or CLI identity and document sharing. Stop at the missing layer instead of broadening access.

3. Run one research item

Create a source record with its link and reading scope. The writer uses that record rather than inventing evidence.

4. Keep a task identity

Assign a task identity, suppress bot echoes and deduplicate events. Verify acceptance and completion separately.

5. Deliver to the test document

Deliver evidence, draft and questions to the test document for human review before production writing or publication.

A synthetic acceptance fixture

handoff.csv: source_url, read_scope, topic_status, draft_doc, reviewer. App reception, CLI identity and document writing each need their own test.

Copyable task brief

Copy template

Use the synthetic fixture to prepare the deliverable for Grok Bot × Feishu: connect X research and drafts through a handoff table. Do not connect real services yet.
Input: Synthetic material: one read public X article and a designated test document.
Requirements:
1. Identify the group, app bot, relay, webhook routine and Grok Bot. Distinguish a group posting webhook from an app that subscribes to messages.
2. Check app scope, user or CLI identity and document sharing. Stop at the missing layer instead of broadening access.
3. Create a source record with its link and reading scope. The writer uses that record rather than inventing evidence.
4. Assign a task identity, suppress bot echoes and deduplicate events. Verify acceptance and completion separately.
5. Deliver evidence, draft and questions to the test document for human review before production writing or publication.
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

A group message traces to one source and one draft, without loops or false success on permission failures.

Copy template

Review the artifacts without taking external actions.
Acceptance: A group message traces to one source and one draft, without loops or false success on permission failures.
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: one coordinator, research handoffs and A/B drafts

The 13.84-second group-chat recording extends the existing Feishu bridge into a visible handoff sequence. The user addresses the coordinator, which assigns research and writing. It shows reviewable drafts and a confirmation prompt, with no successful X publication shown.

Time	Readable action and result
00:00–00:04	Mention the coordinator for recent research; receive a categorized brief, links and possible topics
00:04–00:07	Ask the coordinator to have the researcher investigate the Grok ecosystem
00:07–00:12	Ask the writer to draft from that research; receive short version A and a longer version B with action points
00:12–00:13.8	The ending asks for an A/B choice and separate explicit confirmation before posting to X

Add four steps to the existing handoff table

1. Specify the research window. Give the coordinator a topic, sources and dates; require original links. The recording’s contemporary research is an example from the author’s date, not current reporting.
2. Investigate one selected topic. Keep the task identifier and have the researcher deliver evidence and unknowns. The writer consumes that artifact rather than starting unrelated collection.
3. Deliver A/B drafts separately. Use the same source set, identifying length, angle and unresolved claims. A short version still needs evidence.
4. Record selection and sending separately. Save the chosen version in the ledger; enter a platform action only under an explicit send instruction. The recording has no completed-send receipt, so its A/B selection wording does not establish an X post.

Copy template

Prepare one reviewable coordinator-to-researcher-to-writer handoff for a single topic.
The coordinator lists the topic, date window and authorized sources.
The researcher supplies a linked investigation; the writer produces short A and longer B from it.
List sources and unknowns for each version.
Record task_id, source_record, draft_A, draft_B, selected_version and send_status.
An unselected draft is awaiting selection; a selected draft is still awaiting sending.
This run prepares research and drafts only: no new Bots, bridge changes, messages or publication.

Video verification scope

The collector checked approximately one frame per second, including the first and last frames, and enlarged the ending. The largest viewed-frame gap was one second; this was not continuous viewing. Small research entries were not transcribed individually. An audio track is present, but full-track automatic recognition produced no reliable speech. Whether it is music could not be confirmed by human listening, so this supplement uses readable visuals only. Full role prompts were unavailable and neither the group workflow nor posting was reproduced.

Audio and role-source follow-up: steps still rely on reviewed visuals

The full Feishu audio track underwent another automatic recognition and loudness/silence analysis. The results cannot reliably establish intelligible speech or music only, and speech cannot be excluded. “Thank you for watching” was rejected as narration evidence. There is no human listening confirmation and no new audio-derived procedure.

The post and visible public material still did not provide complete role SOUL.md files, bridge code or a versioned download link. A path, CLI profile or success message in an image does not disclose those files. The existing permission checks and A/B handoff rely on text, images and reviewed video visuals; copyable site prompts remain independently written.

Label handoff evidence as body, image, video visual, audio unconfirmed or site fixture. Keep role-file versions, bridge entry points and their sources in the missing-material list. Unreliable automatic transcripts must not fill in missing roles or configuration.

Audio and source verification status

Audio remains unconfirmed; complete role and bridge sources were not obtained. No installation, authorization, group message or X post was performed, and automatic analysis is not described as human listening.

Second source follow-up: account identified, role and bridge files still missing

The follow-up identified liyue-aigc, whose GitHub profile record associates it with the X username liyue_ai. This supports account attribution; it does not establish access to the Feishu demonstration’s source.

Keep the reading scopes separate: the account’s 14 public repositories were listed and their complete default-branch path trees checked; public file content was searched for SOUL, feishu, x-intel and lark. A path audit is not a full code review. The first three content queries returned no matches. The two lark matches were video-director Skill references to hosted documents, not a messaging bridge. Similar Feishu projects were not substituted.

Full role SOUL.md files, the bridge entry point, role-to-CLI-profile mapping, and actual filtering, deduplication and completion-receipt implementation remain unavailable from the checked sources. Existing steps still use the article and visible demonstration. Any later source delivery needs an original link, pinned version and role-specific paths. Audio remains uncertain, with no new audio-derived steps.

Sources and applicability

Original article: @liyue_ai; published 2026-10-11.

Index-to-source mapping: strong content match.

The collector reports viewing 22 images and reviewing 1 video(s) through sampled frames, without continuous playback. Images add im.message.receive_v1 subscription, IM permissions and app publication. Two CLI profiles separate research and content identities and read their SOUL.md files. Webhook acceptance is not completed delivery.

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

Verification record

Content updated: 2026-10-18 · Editing a page does not reverify its sources or live workflow.

Documentation review: Historical review record · 2026-10-18
Preserves this page’s existing source-review record. Its source notes state actual text, image, video or code coverage and unavailable material. This integration did not recheck external sources or establish a live workflow pass.

Live workflow verification: Unverified
This site has not executed the live accounts, bridges, external writes or scheduled workflow. Author demonstrations, synthetic fixtures and page checkboxes do not establish this site’s live verification.