agentleFS
sunholo-datasunholo-data/ailang/.agents/skills/mission-brief/SKILL.md

mission-brief

Skill38 starsChanged 3 days agoApache-2.0View on GitHub

Read-only steering brief for attended mission sessions — dashboard + directive channel + live loop health + inbox, then hand the wheel to Mark. Use when the user says "mission brief", "catch me up on the mission", "steering session", or opens a fresh thread for in-person steering or meta-loop updates. NOT for running iterations — that is mission-control.

5 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 1

    Distinctive

    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:

    READ-ONLY: this skill gathers and reports.

    line 8
  2. Very specific:

    Dashboard: Read design_docs/mission-dashboard.md (≤40 lines, the full snapshot).

    line 15

“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “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

Mission Brief — the attended steering entry point

READ-ONLY: this skill gathers and reports. No acks, no edits, no commits — steering actions come after the brief, at Mark's direction. mission-control is the machine's entry point into the loop; this is the human's. Never fold brief logic into mission-control (it is live by symlink for every mission the instant it is saved — small blast radius matters).

Gather (all four; they are independent, so run in parallel)

  1. Dashboard: Read design_docs/mission-dashboard.md (≤40 lines, the full snapshot). Staleness check: if its Updated stamp is older than ~2× the v1 cadence on its Loops line (90min default) while the loop claims armed, FLAG IT — Gate-4 isn't refreshing, which is itself a finding (wedged loop, quota dry-out, or launchd off). Normalize to UTC before ANY timestamp comparison: issue comments are UTC (Z suffix), the dashboard stamp is local time. A stamp in the future is itself suspect — the 2026-08-04 bootstrap wrote 14:30Z as "~16:30" via exactly this mixup.
  2. Directives — the channel the dashboard only points at:
    gh api --paginate "repos/sunholo-data/ailang/issues/$(cat ~/.ailang/state/mission-gh-issue)/comments" \
      | jq -rs 'flatten | reverse | .[0:5][] | "\(.created_at) \(.user.login): \(.body | split("\n")[0])"'
    (Last 5 comments, newest first, first line each. --paginate + the standalone jq -s/flatten is load-bearing: the API serves comments oldest-first, 30 per page, so a bare call reads page 1 only — the moment a rotation week crosses 30 comments, "newest 5" goes silently stale. That is exactly how the 2026-08-05 afternoon failure notices went unseen by a brief run at 12:50. gh refuses --slurp with --jq, hence the pipe.) Any comment newer than the dashboard's Updated stamp is an UNPROCESSED DIRECTIVE — surface it verbatim. The pointer file rotates Mondays; if the issue is closed or missing, report that rather than skipping.
  3. Live loop health — the dashboard's "armed" is a snapshot, not now:
    launchctl list | grep -E 'dev\.ailang\.mission-(control|world)'
    Column 1 = PID if running right now, column 2 = last exit status. Then the last logged iteration: grep "^## " design_docs/v1-mission-log.md | tail -1 — compare its date against the cadence, but read the dashboard's quota-posture section before calling a gap "wedged" (a dry-out can be expected posture, not breakage).
  4. Inbox: usually already injected by the SessionStart hook — summarize what it showed, don't re-fetch. Mid-session: ailang messages list --compact. Do NOT ack — Mark's call.

Report

One compact brief, in this order: Now (from dashboard) · unprocessed directives · loop health (live) · inbox highlights · Parked-on-Mark items. Then ask: what do you want to steer?

Steering levers — route Mark's decision through an existing channel, never improvise one

  • Durable directive → comment on the bookkeeping issue (Gate-1 reads it next iteration)
  • Charter change → edit design_docs/v1-mission.md per its attended-stamp conventions
  • Message a loop/agent → ailang messages send (see ailang messages --help)
  • Do it now → this attended session does the work directly
  • Meta/skill fix → edit the skill IN THE MAIN CHECKOUT — a worktree commit reaches origin but never the running loop (mission-control's Gate-5 symlink note has the details)

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.