agentleFS
davekilleendavekilleen/Dex/.agents/skills/meeting-closeout/SKILL.md

meeting-closeout

Skill509 starsChanged 3 months agoMITView on GitHub

Close out the meeting you just had while it's fresh — lock the decisions, the action items and who owns each, what you personally committed to, and the single next step — then capture it and, only with your OK, turn the actions into tracked tasks. Use when the user says 'wrap up this meeting', 'close out my 3pm', 'here are my notes from the call', or right after a meeting ends. Also use proactively when the user pastes raw notes from a meeting that just happened. Not for bulk-processing many already-synced meetings; use `process-meetings`. Not for prepping a meeting that hasn't happened yet; use `meeting-prep`.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If there are no notes and no matching vault note, the search is over — ask for the notes (or a two-line recap) — do not fabricate a summary …

    line 41
  2. Distinctive:

    Never report "captured / done" without those in hand.

    line 70
  3. Distinctive:

    No pasted notes and no matching synced meeting → ask for the notes.

    line 88
  4. Very specific:

    Update System/usage_log.md to mark meeting-closeout as used.

    line 97

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md
<!-- Generated from `.claude/skills/meeting-closeout/SKILL.md` by `scripts/generate-agents-skills.py`. Do not edit. -->

/meeting-closeout

The gap between walking out of a meeting and the next context switch is where decisions and promises evaporate. This closes one meeting the moment it ends — while you still remember who said they'd do what — and locks it down.

It is the single-meeting, in-the-moment ritual. It is not the bulk "catch up all my synced notes" pass (process-meetings) and not pre-meeting prep (meeting-prep).


Step 1 — Get the meeting

Work from whichever exists, in this order:

  1. Notes the user just pasted or dictated — the common case; use them directly even if the meeting was never synced.
  2. The meeting they name ("my 3pm with Acme") — pull it via get_meeting_context.
  3. The user's configured meeting source — read meeting_sources in System/user-profile.yaml. If a notes_folder is set, search that folder for the meeting by title, attendee, and date before anything else. A configured source always outranks improvised discovery.
  4. A provider-neutral source note elsewhere in the vault — if the named meeting was not returned, search the vault by meeting title, attendee, and date. This includes notes created by ClickUp AI and other recorders. Provider-neutral discovery is not only 00-Inbox/Meetings/. Exclude Dex internals, dependency folders, archives that fall outside the requested date, and binary files. Prefer QMD when available, otherwise use a bounded Markdown filename/content search. Read the matched note before treating it as the meeting.

Source boundary (hard rule). Meeting notes come from the vault (and the configured meeting_sources folder) only. Never go looking in external services — Google Drive, Gemini notes, Notion, email, or any connected tool — for a meeting the vault doesn't have, unless the user explicitly points at that source for this meeting in this conversation. Auto-generated notes from an unconfigured source can describe a different meeting or contain invented content, and a closeout built on them is worse than no closeout.

If there are no notes and no matching vault note, the search is over — ask for the notes (or a two-line recap) — do not fabricate a summary of a meeting you can't see, and do not widen the search to external tools.

Step 2 — Extract the closeout essentials

Pull only what the notes support — never invent a decision or an owner the meeting didn't produce:

  • Decisions — what was actually decided (not everything discussed).
  • Action items + owner + due — each task, who owns it, by when. Name the owner explicitly; if the notes don't say, mark it "owner TBD" rather than guessing.
  • Your commitments — what you personally promised (these are the trust-critical ones; they feed the same promise-tracking as commitments).
  • Open questions — unresolved threads to revisit.
  • The single next step — the one thing that moves this forward.

Step 3 — People (respect the entity setting)

Identify attendees and update existing person pages with the relevant context (via lookup_person). For people who don't have a page yet, follow the vault's entity_creation setting — auto creates, suggest (the default) offers, off just tracks. Do not hard-create person pages against the user's setting.

Step 4 — Turn actions into tasks (confirm-gated, never automatic)

Offer to create tasks from the action items and your commitments. Nothing is written without per-item confirmation. For each one the user approves, call Work MCP create_task (carry the owner, due, and the meeting as source; infer the pillar per the CLAUDE.md rules), then read back the created task IDs. Items the user skips are left out; a failed create_task is reported as not created, never counted as done.

Step 5 — Capture, then confirm what happened

Save the closeout to the original source note when one was discovered, wherever it lives. Do not silently copy a ClickUp or other provider note into 00-Inbox/Meetings/. Use 00-Inbox/Meetings/ only when the notes were pasted or dictated and have no source file. Then confirm the real result by reading it back: the note path saved, the tasks created (by ID), and which person pages were updated. Never report "captured / done" without those in hand.


Quality bar

A good closeout leaves the meeting's decisions, owned actions, your commitments, and one next step captured while they're fresh — each grounded in what was actually said — so nothing important is lost between the meeting and the next thing. Owners are named or honestly TBD; nothing is created the user didn't confirm.

Anti-patterns (do not do these)

  • Re-running the bulk sync/catch-up over many meetings — that's process-meetings; closeout is this one meeting.
  • Inventing decisions, owners, or commitments the notes don't support — mark unknowns TBD instead.
  • Auto-creating tasks without per-item confirmation, or person pages against the entity_creation setting.
  • Claiming "captured / done" without reading back the saved note path and created task IDs.
  • Dumping the whole transcript back as "notes" instead of extracting decisions and owned actions.

Degradation

  • No pasted notes and no matching synced meeting → ask for the notes; never fabricate a recap.
  • entity_creation: off → track people, don't create pages; suggest → offer, don't auto-create.
  • A create_task call fails → report that item as not created; don't count it.
  • QMD/semantic search absent → link people/projects by keyword; the closeout still runs.

Track Usage (Silent)

Update System/usage_log.md to mark meeting-closeout as used. Analytics (Silent): call track_event with event_name meeting_closed_out and properties tasks_created and decisions_captured (counts only — no meeting content, no names). Fires only if the user opted into analytics; no action if it returns "analytics_disabled".

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.