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

ceo

Skill4 starsChanged 5 months agoMITView on GitHub

Work the board like the CEO/CTO — read the ENTIRE kanban, make an educated decision on every open card (do now / unblock / wait / human / stale), then EXECUTE the highest-leverage move immediately, invoking any other skill or tool the move needs. Consume-first — drain existing cards before creating new GTM cards or building anything. Use when the owner says "/ceo", "what's the banger move (and do it)", "work the board", "read the kanban and do something", or an interactive session should act instead of report.

10 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use cpo and cmo.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • sahinSkill

    cpo

    CPO Co-Pilot - product strategy, roadmap prioritization, competitive positioning, and product-market fit validation

    Also in 3 projects that use this

  • sahinSkill

    cmo

    CMO Co-Pilot - GTM strategy, growth frameworks, and marketing leadership (composite GTM leader voice)

    Also in 4 projects that use this

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 0 are distinctive, and 4 are very specific to this project.

  • 0

    Common

    Most of the 9 edited versions we read keep these.

  • 0

    Distinctive

    This author's own choices. Worth a look.

  • 4

    Very specific

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

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

    Every card WAIT or HUMAN, nothing DO/UNBLOCK/DUE?

    line 64
  2. Very specific:

    Every card touched gets its status + a human-glanceable note updated via tools/tasks.sh update (5-second-readable, "Your move:" section — …

    line 70
  3. Very specific:

    Verify what you claim done against its success criterion (#13).

    line 72
  4. Very specific:

    Run the close-out audit (tools/done-check.md).

    line 75

“Common” is measured: kept by at least 6 in 10 of the 9 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

ceo — decide on everything, then do the banger move

One interactive pass that ends in action, not a menu. Obey CLAUDE.md throughout (all principles + hard rules). This is NOT a tick (autocomp-tick owns the heartbeat, ledger cadence, and registry) — it's the co-founder sitting down, reading the whole board, judging every card, and executing the single best move right now.

The consume-first law (the owner's directive, 2026-07-07): while ANY existing card is actionable by the loop, the banger move comes from the board. No new GTM cards, no new pages/features/ventures, no fresh ideation — creation is only allowed once every open card is genuinely waiting on a date, the human, or an external queue. Draining beats adding; the board is the backlog we already chose.

1. Read everything (never from memory)

  • tools/tasks.sh list-all — EVERY open card: todo/doing/blocked, agent AND human.
  • tools/agent-task.sh workers — the home-browser queue: what's queued/claimed/done, and whether a worker has claimed anything recently (unclaimed queued jobs = worker down).
  • private/state/approvals.md — any PENDING the owner may have answered (also check - [y]/[n] verdict lines on approval cards via tools/tasks.sh get <id>).
  • Today's date vs. every "due/re-poll after <date>" in card notes.
  • tools/scoreboard.sh — only if you reach §4 (a new pick needs the funnel truth).

2. Judge every card — one verdict each, no card skipped

Assign each open card exactly one verdict:

VerdictMeaningWhat you do with it
DOLoop-doable now with the keys/tools we holdCandidate for §3 — execute
UNBLOCKStuck on a dependency we can fix (queue a job, install, provision, host a file)Fix the dependency now — a self-fixable blocker is not a blocker (#15)
DUEDate-gated and the date has arrivedTreat as DO
WAITDate-gated (future) or sitting in an external review queueLeave; note the date it re-enters play
HUMANIrreducible owner action (their identity / spend / send / judgment)Leave on board; if any sub-step became loop-doable since it was written, do that part and update the card (#12)
STALEOvertaken by events, duplicate, or wrongly-scopedUpdate or close it with an honest note — a rotting card poisons triage

Judge honestly: a card parked as WAIT/HUMAN that has a loop-doable half is really UNBLOCK. That misclassification is the timid failure mode this skill exists to kill.

3. Pick the banger move — and execute it

Rank the DO + UNBLOCK + DUE set: binding-constraint first, then the Hundred-Dollar Test (#6), then whatever other cards depend on. Then do it — now, in this session, and keep chaining through the ranked set as long as the session has room (#18: finish whole tasks, don't stop at seams).

Invoke whatever the move needs — skills are tools, route freely:

  • bot-walled page → fetch-protected; interactive logged-in web task → queue via tools/agent-task.sh (and if the home worker is down, homelab to get it up — that's UNBLOCK);
  • a fact you lack → deep-research / web search; a deploy/DNS/DB step → the matching tools/*.md;
  • anything else with a matching skill → use it rather than hand-rolling.

Hard rules still hard: money / owner's personal identity / mass-cold send / destructive → private/state/approvals.md as PENDING + tools/notify.sh push, never executed here. Bot-identity GTM is pre-authorized — just do it.

4. Only if the board is fully drained

Every card WAIT or HUMAN, nothing DO/UNBLOCK/DUE? Then — and only then — invoke next-move for ONE new pick (it runs the Rule-17 distribution-first gate itself), execute its buildable half, and board the remainder. One pick, not a brainstorm.

5. Close out (before you stop)

  • Every card touched gets its status + a human-glanceable note updated via tools/tasks.sh update (5-second-readable, "Your move:" section — CLAUDE.md #12).
  • Verify what you claim done against its success criterion (#13); spawn a fresh verifier subagent for anything non-trivial (tools/verify-goal.md).
  • Append what happened to private/state/ledger.md; durable learnings → private/memory/.
  • Run the close-out audit (tools/done-check.md).
  • Report: one line per card — verdict → what changed — then the move(s) executed with evidence, then what you did NOT do (#11).

Anti-patterns

  • Building while the board has open DO cards — the exact tar pit the consume-first law bans.
  • Reporting a triage instead of acting — the output of /ceo is changed state, not a table.
  • "Worker is down" filed as WAIT — if this box (or homelab) can start it, it's UNBLOCK.
  • Creating a new card to avoid an old one — drain, don't reshuffle.

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.