playbook
Skill84 starsChanged 5 days agoMITView on GitHub
Orchestrator playbook — the full delegation contract (research pipeline, worker report contract, spawn economics, forks, teammate lifecycle, chair hygiene). The chair MUST load this before its first delegation of every session; the injected core profile only summarizes it.
17 other projects use a version of this.
Adopted by
Most recently taken up by
- n1kben/agentstheir own version, first seen Nov 16, 2026
- Rylaa/fable5.2-opus5.5-orchestratorword for word, first seen Nov 15, 2026
- Rylaa/fable5-opus5-orchestratorword for word, first seen Nov 11, 2026
- SHYXIN/skillstheir own version, first seen Nov 1, 2026
- Hainrixz/whatsapp-closer-agentkittheir own version, first seen Oct 25, 2026
- RumbleJack56/skillstheir own version, first seen Oct 20, 2026
Other versions17 projects use a version of this.
Word for word · 2
- Rylaa/fable5-opus5-orchestrator★ 84100% same
- Rylaa/fable5.2-opus5.5-orchestrator★ 84100% same
Edited · 0
None found.
Wrote their own version · 15
- Hainrixz/whatsapp-closer-agentkit★ 39same name
- A/claude-booping★ 15same name
- SHYXIN/skills★ 6same name
- nauvalazhar/skills★ 4same name
- ThilinaTLM/agent-skills★ 4same name
- ankitai-s/superdialog-playbook-skill★ 2same name
- healthier-vitamins/agentic-grimoire★ 2same name
- roethlar/AgentGovernanceBootstrap★ 2same name
- RumbleJack56/skills★ 2same name
- modeled-information-format/mif-docs-plugin★ 1same name
5 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
We split this skill into the rules it gives an agent
2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.
0
Common
Most of the 10 edited versions we read keep these.
2
Distinctive
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 26
If a choice would change the work and only the user can settle it, ask the USER before spawning.
- Distinctive: line 58
Every spawn pays a fixed overhead (system prompt, project rules, tool schemas) before doing any useful work.
“Common” is measured: kept by at least 6 in 10 of the 10 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.
Files
Orchestrator Playbook
Applies to both chair profiles (FABLE and OPUS). The injected core profile always wins on routing and limits; this file is the detail behind its one-liners.
Writing a worker's brief
A worker cannot ask you anything. Every ambiguity you leave in a spawn prompt becomes a guess committed to code, so the brief carries what the worker would otherwise have to invent:
- the goal in one sentence, and what "done" looks like
- the files or areas it owns, and what it must NOT touch
- constraints that are not visible in the code (a contract another team parses, a deadline, a decision already made)
- what to do when it hits something unexpected: report and stop, or decide and note it
- the exact deliverable — a diff, a brief, a path
If a choice would change the work and only the user can settle it, ask the USER before spawning. One question, in the same message as whatever else you can get done without the answer.
Research pipeline — parallel fan-out, no mid-flight dumps
YOU pick the questions and sources — never a fetch worker. ONE
sonnet (medium) per source: it fetches the source VERBATIM to
./.workflow/scratch/ FIRST (the disk copy is the audit trail — no
relevance filtering during fetch), THEN returns a brief built from
that disk copy: claims, evidence, exact quotes, confidence,
contradictions, and the path. A final sonnet (high) synthesizes
across the briefs. YOU check the synthesis and its verbatim evidence
and decide. Intermediates never enter your context.
Worker report contract (enforced)
Every worker returns:
- what it did, against the brief
- summary
- VERBATIM code/config/errors/quotes the conclusion depends on — at most 10 lines inline; anything longer goes to ./.workflow/scratch/ and the report carries the path
- confidence: "confident" / "uncertain because X"
- "out of scope but noticed"
Reports are at most 40 lines TOTAL. A violating return is rejected and re-run — never silently accepted.
Spawn economics — batch before you multiply
Every spawn pays a fixed overhead (system prompt, project rules,
tool schemas) before doing any useful work. Batch similar mechanical
steps into ONE worker with a checklist; spawn separately only when
true parallelism or isolation pays for that overhead. Many similar
items: run them at low and re-run only the failures at high. Read-only
workers share the repo concurrently; parallel EDITORS each run with
isolation: "worktree".
Forks
subagent_type: "fork" clones your FULL conversation at your model
and spends the usage limit: at most 2 per session, only while the
conversation is still short, and only for bounded follow-ups that
lean on context a spec cannot carry. Forking a plan's phases is
disguised solo work — phases go to workers with specs.
Named teammates — the user watches the work
NAME every substantive worker (implementation, review, research):
named teammates run in tmux panes the user watches live, and their
lifecycle states reach the chat; an unnamed subagent is a silent
spinner until it returns. Only sub-minute lookups (a grep, one
read/fetch) stay unnamed. Steer a running teammate mid-task with
SendMessage. Once its final report is ACCEPTED with no follow-up
planned, dismiss it: SendMessage {"type": "shutdown_request"}.
Dismissal is final, so dismiss only after processing the output —
and never leave finished teammates stacked (the plugin reaps
forgotten panes).
Checking the work
You accept or reject every report; that is the arbiter's job and it is not delegated. When a worker's own testimony is the only evidence that its work is right — it wrote the code AND declared it passing — get the evidence from somewhere else: run the tests yourself, read the diff, or send a FRESH worker that did not build it. Scale that to the blast radius, not to every close: a one-file mechanical edit does not need a second opinion, an irreversible migration does.
Chair context hygiene
Consume briefs + verbatim snippets; bulk stays on disk. When a decision hinges on exact content that is short, read it yourself — never decide on a summary when the source fits in a few hundred lines. Prefer per-task sessions: scratch survives /clear, so finish a task, close it, start the next one clean. Drop closed-phase raw material; keep outputs minimal; parallelize independent calls.
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
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.

