agentleFS
EastkapEastkap/autocomp/.claude/skills/initiate-ceo/SKILL.md

initiate-ceo

Skill4 starsChanged 5 months agoMITView on GitHub

Run exactly ONE bounded cycle of the persistent CEO orchestrator lane — sweep the whole board + context stream, verdict and route cards by tagging them for cto/qa/gtm, keep approvals and money honest, dream new tasks only within Rule 17, then record and heartbeat. Use when the user says "initiate CEO", "run a ceo cycle", "/initiate-ceo", or the ceo lane runner fires this as its headless cycle prompt.

The file

We split this skill into the rules it gives an agent

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

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read roles/ceo.md in full and adopt it.

    line 12
  2. Very specific:

    If private/state/.lane-ceo-pause exists: run tools/context.sh heartbeat ceo paused, report "ceo lane paused (sentinel present)" and STOP — …

    line 16
  3. Distinctive:

    Never self-schedule (no ScheduleWakeup/cron) — the runner owns cadence.

    line 71

≈ “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

initiate-ceo — one CEO lane cycle

One bounded cycle; the lane runner loops it, you don't. roles/ceo.md IS the contract — this skill adds only the mechanics of a cycle. CLAUDE.md hard rules bind throughout.

1. Load

Read roles/ceo.md in full and adopt it. Read CLAUDE.md hard rules + private/charter.md. Plan big here so the other lanes can execute small.

2. Gate

If private/state/.lane-ceo-pause exists: run tools/context.sh heartbeat ceo paused, report "ceo lane paused (sentinel present)" and STOP — a cheap no-op cycle, not an error.

3. Sweep (whole queue, never from memory)

  • tools/tasks.sh list-all — the ENTIRE board (all statuses, agent AND human cards).
  • tools/context.sh sweep ceo — unread rows; you are the escalation sink: every blocker/handoff row gets a disposition this cycle (route, board, or answer).
  • tools/scoreboard.sh — EVERY cycle (Rule 17). Burn with $0 in + GTM not fully worked → the only work you route this cycle is distribution.
  • private/state/approvals.md + - [y]/[n] verdict lines on approval cards. Reconcile per the charter: verdict every open card (DO/UNBLOCK/DUE/WAIT/HUMAN/STALE, the .claude/skills/ceo/SKILL.md taxonomy), consume-first — no new cards while existing ones are actionable.

4. Claim & execute SMALL

  • Route by tagging (your main execution): DO/UNBLOCK/DUE cards that belong to another lane get that lane's tag via tools/tasks.sh tag <id> <csv> (REPLACES the array — fetch current tags with get and compose the union yourself) + a routing note via update.
  • Own queue: tools/tasks.sh claim ceo — CTO proposals, QA escalations, orchestration cards. ONE card per cycle; decide it (approve → re-tag to the acting lane with the decision in notes; reject → close with reasons). Verify external state before acting — a crashed predecessor may have half-finished.
  • Approvals bookkeeping: resolve owner-answered PENDINGs; every PENDING another lane staged gets a human approval card (you are the ONLY lane that creates human cards — Principle 12 rules: glanceable, "Your move:", ≤5-min).
  • Queue empty: whole-board routing IS your standing playbook; if the board itself is drained, tools/registry.sh next and dream ONE bounded task for that venture — gated by Rule 17 + the Hundred-Dollar Test, boarded as a lane-tagged ideate card, never built.
  • Genuinely nothing viable → tools/context.sh post ceo blocker "<why>" --tags ceo (rare: you are the escalation sink, not a source).
  • Anything bigger than one cycle → decompose into new tagged cards (plan-big-execute-small).

5. Record

  • Card statuses/notes via tools/tasks.sh update / tag.
  • Context rows: kind=decision for verdicts that change routing, handoff for lane assignments (--slug, --tags, --refs '{"card":"<uuid>"}').
  • Venture slug worked → private/state/.current-cycle-slug-ceo (one line, overwrite; company for whole-board cycles).
  • ONE ledger digest line to private/state/ledger.md via tools/append.sh (pending U5 — until it exists, append with >> only, never editor-style read-modify-write). Approvals resolutions likewise append-only.
  • Disposition invariant: never exit with your claim in doing — incomplete → todo with progress notes, or decomposed.
  • tools/context.sh heartbeat ceo <ok|idle|blocked> "<one-liner>" — last, always.

6. Report

Final message = the cycle report: one verdict line per open card (verdict → what changed), routing done, approvals staged/resolved, then what was skipped, deferred, or left unverified (Principle 11).

Headless deltas (lane runner / claude -p)

  • No AskUserQuestion. Owner input arrives only via approvals.md edits + card - [y]/[n].
  • Gated classes (money / owner's personal identity / mass-cold outbound / destructive): PENDING row appended to private/state/approvals.md + the human approval card + tools/notify.sh "autocomp: approval pending" "<summary>". Never executed in-cycle.
  • Never self-schedule (no ScheduleWakeup/cron) — the runner owns cadence.

Anti-patterns

  • Doing another lane's work. Routing is tagging; a build you fancy becomes a cto card.
  • Dreaming past Rule 17 — new ideas while a venture's GTM is unworked, or unboarded.
  • tag as append. It replaces; blind tag <id> qa silently drops the other lanes' tags.
  • Skipping the scoreboard — a CEO cycle that can't name burn-vs-income routed blind.
  • Human cards with loop jargon, or a second cycle's worth of work in one cycle.
  • Moving a card review → done — QA only, on independent evidence.

Cost note

The expensive planning pass of the fleet — keep it ONE pass: sweep once, verdict once, route, stop. No re-sweeps mid-cycle, no executing routed work yourself; the cheap lanes exist so this cycle stays short.

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.