/fable-ish
Apply fable-ish verification discipline to Claude Code coding tasks with risk-proportional workflow, review, and final reporting. Use for fable-ish requests, common typo fablish, complex implementation, debugging, refactoring, deployment preparation, or work where completion
$ npx -y skills add chrisryugj/fable-ish --skill fable-ish --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-ish
Context preview
The summary Claude sees to decide when to auto-load this skill.
Apply fable-ish verification discipline to Claude Code coding tasks with risk-proportional workflow, review, and final reporting. Use for fable-ish requests, common typo fablish, complex implementation, debugging, refactoring, deployment preparation, or work where completion
SKILL.md
fable-ish.SKILL.mdname: fable-ish
description: Apply fable-ish verification discipline to Claude Code coding tasks with risk-proportional workflow, review, and final reporting. Use for fable-ish requests, common typo fablish, complex implementation, debugging, refactoring, deployment preparation, or work where completion needs observable evidence. Do not use for simple explanations or quick low-risk answers unless explicitly requested.
fable-ish
Use this skill as the workflow instruction layer for the fable-ish plugin. The plugin hooks provide mechanical guardrails and a verification gate (task classification, risky-action blocking, evidence tracking, stop-time review); this skill provides the human-readable workflow. Treat `fablish` as a common typo alias for `fable-ish`; do not rename the plugin, skill, or directory.
Core Rule
Match verification depth to task risk.
- Quick tasks: answer directly and do not force deep planning.
- Normal coding tasks: inspect relevant context, make a coherent change, and run one relevant proof when files change.
- Deep tasks: define the work unit and observable exit proof before final response.
- Blocked tasks: stop when the next action needs user confirmation, credentials, external state, or unsafe destructive scope.
Preserve domain meaning. Treat user-provided rules, edge cases, constraints, and correction criteria as primary inputs for implementation and verification.
Do not claim verification that was not observed.
References
Read only the reference needed for the current task:
- `references/workflow.md`: mode selection, work-unit loop, optional review passes.
- `references/verification.md`: choosing and reporting proof.
- `references/final-report.md`: concise final report format.
Operating Notes
- Keep small requests small.
- Prefer existing repo commands, tests, validators, and docs.
- Use the task's work kind to choose proof, not only code-test defaults.
- If no verifier exists, create a small one only when it is low-risk and clearly useful.
- Use subagents only when they increase recall for broad or risky work; they are optional, not mandatory.
- Stop at a verified boundary or name the concrete remaining blocker.
Read more
name: fable-ish description: Apply fable-ish verification discipline to Claude Code coding tasks with risk-proportional workflow, review, and final reporting. Use for fable-ish requests, common typo fablish, complex implementation, debugging, refactoring, deployment preparation, or work where completion needs observable evidence. Do not use for simple explanations or quick low-risk answers unless explicitly requested.
fable-ish
Use this skill as the workflow instruction layer for the fable-ish plugin. The plugin hooks provide mechanical guardrails and a verification gate (task classification, risky-action blocking, evidence tracking, stop-time review); this skill provides the human-readable workflow. Treat `fablish` as a common typo alias for `fable-ish`; do not rename the plugin, skill, or directory.
Core Rule
Match verification depth to task risk.
- Quick tasks: answer directly and do not force deep planning.
- Normal coding tasks: inspect relevant context, make a coherent change, and run one relevant proof when files change.
- Deep tasks: define the work unit and observable exit proof before final response.
- Blocked tasks: stop when the next action needs user confirmation, credentials, external state, or unsafe destructive scope.
Preserve domain meaning. Treat user-provided rules, edge cases, constraints, and correction criteria as primary inputs for implementation and verification.
Do not claim verification that was not observed.
References
Read only the reference needed for the current task:
- `references/workflow.md`: mode selection, work-unit loop, optional review passes.
- `references/verification.md`: choosing and reporting proof.
- `references/final-report.md`: concise final report format.
Operating Notes
- Keep small requests small.
- Prefer existing repo commands, tests, validators, and docs.
- Use the task's work kind to choose proof, not only code-test defaults.
- If no verifier exists, create a small one only when it is low-risk and clearly useful.
- Use subagents only when they increase recall for broad or risky work; they are optional, not mandatory.
- Stop at a verified boundary or name the concrete remaining blocker.
"fable-ish" = "Fable처럼(-ish)". 관찰 가능한 검증으로 통과하기 전에는, 완료를 인정하지 않는다. fable-ish는 Anthropic이 최근 선보인 코딩 모델 Fable이 보여준 검증 규율을 Claude Code 훅으로 재현한 플러그인입니다.
Repo: chrisryugj/fable-ish