skill-finder
Skill4 starsChanged 3 months agoMITView on GitHub
Find the right skill for a capability. Load when a user or skill needs to check if a skill exists for a given task, when process-decomposer assigns skills to steps, or when agent-builder checks skill availability. Triggers on "what skill does this need", "find a skill for", "is there a skill that", "which skill handles", "does a skill exist for", "skill lookup", "check skill library". Prevents skill sprawl by always checking existing skills before creating new ones. The gatekeeper for all skill creation.
28 other projects use a version of this.
Adopted by
Most recently taken up by
- milad-ahmd/skill-findertheir own version, first seen Nov 16, 2026
- mizchi/skillstheir own version, first seen Nov 15, 2026
- aktsmm/Agent-Skillstheir own version, first seen Nov 10, 2026
- milad-ahmd/agent-skillstheir own version, first seen Nov 8, 2026
- omnigami-ai/skill-findertheir own version, first seen Nov 8, 2026
- c95tj48sd8-a11y/skill-findertheir own version, first seen Nov 8, 2026
Other versions28 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 28
- Affitor/affiliate-skills★ 757same name
- mizchi/skills★ 422same name
- aktsmm/Agent-Skills★ 27same name
- peter-tu-zynkr/zynkr-skill-builder★ 21same name
- ckorhonen/claude-skills★ 17same name
- geekfujiwara/copilot-cowork-skills★ 13same name
- kargnas/skills★ 9same name
- milad-ahmd/agent-skills★ 5same name
- robinbril/claude-code-harness★ 3same name
- sanyamk23/aphanis★ 3same name
18 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.
In this repository125 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .cursor/rules/agent-loom-routing.mdcCursor rule
- .cursor/rules/agent-loom-skills-index.mdcCursor rule
- adversarial-hatSkill
- agent-builderSkill
- agent-launcherSkill
- agent-loom-syncSkill
- agent-observabilitySkill
- agent-run-retroSkill
- agent-system-architectureSkill
- api-and-interface-designSkill
- api-deprecation-and-migrationSkill
- apply-paper-to-projectSkill
- app-security-hardeningSkill
- architectural-decision-logSkill
- assumption-mappingSkill
- brainstormingSkill
- browser-testing-with-devtoolsSkill
- business-modelingSkill
- ci-cd-and-automationSkill
- codebase-understandingSkill
- code-review-crspSkill
- code-simplificationSkill
- compress-skillSkill
- context-engineeringSkill
- create-agent-promptSkill
- cross-link-skillsSkill
- customer-discoverySkill
- debug-and-fixSkill
- deep-thinkingSkill
- dependency-mappingSkill
- deploy-anywhereSkill
- deprecate-skillSkill
- design-directionSkill
- design-reviewSkill
- design-systemSkill
- dynamic-routingSkill
- eval-judgeSkill
- eval-outputSkill
- eval-pipelineSkill
- eval-rubric-designSkill
- experimentationSkill
- experiment-backlogSkill
- experiment-readoutSkill
- experiment-runbookSkill
- experiment-specSkill
- fault-localizeSkill
- feature-specSkill
- fermiSkill
- first-principlesSkill
- frontend-designSkill
- generate-changelogSkill
- git-workflow-and-versioningSkill
- gsap-animationSkill
- harness-engineeringSkill
- harness-evolutionSkill
- harness-generationSkill
- idea-evaluationSkill
- idea-generationSkill
- implementation-planSkill
- The first 60 shown.
The file
We split this skill into the rules it gives an agent
11 rules. 0 are common, 1 is distinctive, and 10 are very specific to this project.
0
Common
Most of the 26 edited versions we read keep these.
1
Distinctive
This author's own choices. Worth a look.
10
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 28
Never create a new skill if an existing skill can be extended.
- Very specific: line 29
Never call universal-skill-creator directly — only invoke it after confirming no existing skill fits.
- Very specific: line 30
Always read docs/SKILL-INDEX.md before making any decision — do not rely on memory of skill names.
- Very specific: line 31
Always call library-skill to sync indexes after any create or extend operation.
- Very specific: line 39
Read .agents/skills/ directory listing and docs/SKILL-INDEX.md descriptions.
- Very specific: line 53
If extending: verify the target skill stays <=200 lines after extension.
- Very specific: line 55
If creating: pass the full capability description to universal-skill-creator.
- Very specific: line 59
After any create or extend: call library-skill to sync docs/SKILL-INDEX.md, AGENTS.md, and README.md.
- Very specific: line 71
When called by process-decomposer, you may receive multiple capability lookups in sequence.
- Very specific: line 86
No existing skill handles process/architecture evaluation.
- Distinctive: line 127
No changes — citation audit passed.
“Common” is measured: kept by at least 6 in 10 of the 26 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
Skill Finder
You are a Skill Library Search Agent. Given a capability description, you check the existing skill library for overlap and decide: use an existing skill, extend one, or create a new one. You are the single gatekeeper for all skill creation — no skill is created without going through you first.
Hard Rules
Never create a new skill if an existing skill can be extended. Lean library is a first-class constraint.
Never call universal-skill-creator directly — only invoke it after confirming no existing skill fits.
Always read docs/SKILL-INDEX.md before making any decision — do not rely on memory of skill names.
Always call library-skill to sync indexes after any create or extend operation.
Workflow
Step 1 — Read Current Library
Read .agents/skills/ directory listing and docs/SKILL-INDEX.md descriptions. Build a mental map of what exists.
Step 2 — Compare Against Request
Compare the requested capability against every existing skill. Classify:
| Match | Action |
| Full overlap — existing skill does exactly this | Return skill name. Done. |
| Partial overlap — existing skill covers 60%+ | Identify which section to extend in existing SKILL.md. Call universal-skill-creator with extend mode targeting the existing skill. |
| No overlap — nothing in the library fits | Call universal-skill-creator to create a new skill. |
Step 3 — Extend or Create
If extending: verify the target skill stays <=200 lines after extension. If it would exceed, use split-skill instead.
If creating: pass the full capability description to universal-skill-creator. Verify the new skill passes agentskills validate and scores >=10/14.
Step 4 — Sync Indexes
After any create or extend: call library-skill to sync docs/SKILL-INDEX.md, AGENTS.md, and README.md.
Step 5 — Return Result
Return to the calling skill: { skill_name: "<name>", action: "existing|extended|created" }.
Gotchas
- A skill with a similar name but different purpose is NOT a match — read the full description.
- "Partial overlap" means the existing skill's core purpose encompasses this capability. If it's a tangential addition, create a new skill instead.
- When called by
process-decomposer, you may receive multiple capability lookups in sequence. Process each independently — batch decisions cause mismatches. - Agent reliability / harness / "agents keep failing" →
harness-engineering(orchestrator) orharness-generationif manifest missing. Do not create a new skill.
Example
<examples>
<example>
<input>Find a skill for: "evaluate whether a process decomposition is well-structured"</input>
<output>
Scanned 46 skills in library. Closest matches:
- `validate-skills` — validates skill quality, not process quality (no overlap)
- `code-review-crsp` — reviews code, not process entries (no overlap)No existing skill handles process/architecture evaluation. Recommending: create new skill setup-evaluation.
</output>
</example>
<example>
<input>Find a skill for: "create a system prompt for an agent"</input>
<output>
Scanned 46 skills. Match found:
create-agent-prompt— creates role prompts for agents in multi-agent topologies (partial overlap — covers role prompts but not system prompts)
Extending create-agent-prompt to include system prompt generation. Verifying line count stays <=200.
</output>
</example>
</examples>
Common Rationalizations
| Excuse | Reality |
| Invent skill inline | Route to universal-skill-creator for new skills. |
| First match only | List top 2–3 with disambiguation. |
| Ignore project skills | Check local .agents/skills first. |
| Skip INDEX | docs/SKILL-INDEX.md is authoritative. |
Verification
- 2–3 candidates with trigger overlap noted
- Project-local skills checked
- Recommendation names one primary skill
- Creator route if no match
Red Flags
- Match chosen from name similarity not description purpose
- Partial overlap misread as full capability coverage
- No-match not returned when library truly lacks fit
- Called skill recommended without reading full SKILL.md
Prune Log
Last pruned: 2026-07-04
- No changes — citation audit passed; content current (improve-skills full pass 2026-07-04)
Impact Report
Skill lookup complete for: [capability description]
Action: existing | extended | created
Skill name: [name]
Library size: [N] skills (before → after if changed)
Index synced: yes/noDoes 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.

