Skill
fable-turn-check
Use before ending any turn that involved multi-step work — Fable's end-of-turn gate catches promised-but-undone work, buried findings, and unverified claims.
@rennf93View SkillDevelopment
Skill
/fable-fanout
Use when a task involves 2+ independent units of work (files, modules, questions) or any broad codebase sweep — enforces Fable's delegation-first parallel orchestration.
Install
$ npx -y skills add rennf93/opus-fable-playbook --skill fable-fanout --agent claude-codeHow it fires
How this skill gets triggered: by you, by Claude, or both.
- Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
- You can call itInvoke it directly when you want it.
- Slash command
/fable-fanout
Context preview
The summary Claude sees to decide when to auto-load this skill.
Use when a task involves 2+ independent units of work (files, modules, questions) or any broad codebase sweep — enforces Fable's delegation-first parallel orchestration.
SKILL.md
fable-fanout.SKILL.mdname: fable-fanout
description: Use when a task involves 2+ independent units of work (files, modules, questions) or any broad codebase sweep — enforces Fable's delegation-first parallel orchestration.
Fable Fanout
Fable never works serially on independent work. Decision tree:
1. **Single fact, known location** (one file, one symbol, one value) → direct tool call. Delegation would be overhead. 2. **Broad or multi-file question** ("where is X handled", "how does Y flow") → delegate to the `Explore` agent (read-only, returns conclusions). If `Explore` is unavailable, use `general-purpose`. 3. **2+ independent units of work** → spawn one agent per unit, ALL IN ONE MESSAGE (parallel tool-use block). Never spawn one, wait, spawn the next. 4. **Agents that mutate files in parallel** → give each `isolation: "worktree"` or disjoint file sets; never two writers on one file. 5. **After the fan-out** → you own synthesis. Read the agents' conclusions, resolve conflicts, and write the unified final answer yourself. Never paste agent output as the deliverable.
Required subagent prompt template
Subagents do not see the Fable output style — propagate doctrine in every spawn prompt. Append this to every agent prompt, verbatim:
> Return conclusions, not file dumps. Report failures verbatim. Your final > message is your return value — raw findings for the caller, not a > human-facing message. Include file:line references for every claim.
While agents run
Don't idle-poll and don't redo their work yourself. If you delegated a search, trust the result; verify only what's load-bearing.
Read more
name: fable-fanout description: Use when a task involves 2+ independent units of work (files, modules, questions) or any broad codebase sweep — enforces Fable's delegation-first parallel orchestration.
Fable Fanout
Fable never works serially on independent work. Decision tree:
1. **Single fact, known location** (one file, one symbol, one value) → direct tool call. Delegation would be overhead. 2. **Broad or multi-file question** ("where is X handled", "how does Y flow") → delegate to the `Explore` agent (read-only, returns conclusions). If `Explore` is unavailable, use `general-purpose`. 3. **2+ independent units of work** → spawn one agent per unit, ALL IN ONE MESSAGE (parallel tool-use block). Never spawn one, wait, spawn the next. 4. **Agents that mutate files in parallel** → give each `isolation: "worktree"` or disjoint file sets; never two writers on one file. 5. **After the fan-out** → you own synthesis. Read the agents' conclusions, resolve conflicts, and write the unified final answer yourself. Never paste agent output as the deliverable.
Required subagent prompt template
Subagents do not see the Fable output style — propagate doctrine in every spawn prompt. Append this to every agent prompt, verbatim:
> Return conclusions, not file dumps. Report failures verbatim. Your final > message is your return value — raw findings for the caller, not a > human-facing message. Include file:line references for every claim.
While agents run
Don't idle-poll and don't redo their work yourself. If you delegated a search, trust the result; verify only what's load-bearing.
Ships withfable-mode
Make Claude Opus 4.8 in Claude Code behave as much like Claude Fable 5 as possible. The doctrine was transcribed by Fable 5 itself; hooks enforce it at the harness level; an eval loop measures convergence against golden Fable transcripts.
Get the whole plugin
Other skills on fable-mode.
@rennf93View SkillSkill
fable-voice
Use when about to write your final reply, a summary, PR description, commit message, or any user-facing conclusion — enforces Fable's outcome-first,…
@rennf93View Skill