agentleFS
Sign inStart free
matthullmatthull/my-skills/skills/workflow/skill-discovery/SKILL.md

skill-discovery

Skill0 starsChanged 55 days agoMITView on GitHub

Discover and load relevant skills for the current task. Scans the installed skill catalog, evaluates relevance, and outputs reasoning. Use at the start of any orchestrated agent work to ensure the right practice skills are loaded.

We split this skill into the rules it gives an agent

4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 12 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every skill gets a YES/NO with a brief justification.

    line 40
  2. Very specific:

    Read resources/domain-skill.md if it exists in the project's .claude/skills/ directory.

    line 44
  3. Distinctive:

    If no skill matches what you need:

    line 52
  4. Distinctive:

    Do not stall because a skill is missing.

    line 57

“Common” is measured: kept by at least 6 in 10 of the 12 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

Skill Discovery

Find which installed skills are relevant to your current task, load them, and document your reasoning. Run this before doing substantive work.

When to Use

At the start of any task that might benefit from practice skills — implementation, design, research, QA, or orchestration. If you are unsure whether skills exist for your task, run discovery anyway.

Discovery Process

1. Locate the Skill Catalog

Your system context includes a catalog of all installed skills. Look for the block listing available skills in the Skill tool definition or the <available-skills> section. This is your source of truth — do not search the filesystem.

2. Evaluate Each Skill

For every skill in the catalog, read its name and one-line description. Decide: is it relevant to this task's technology, domain, or workflow?

3. Output Reasoning (MANDATORY)

You MUST output your selection reasoning for transparency and debugging. Format:

Skill discovery:
- supabase-sql: NO — no database policy work in this task
- react-native-expo: NO — this is a Rails backend task
- pg-perf-power-user: YES — task involves query optimization
- lookup-docs: YES — task references unfamiliar library API
- ruby-rails: YES — Rails service object implementation

Every skill gets a YES/NO with a brief justification. Do not skip skills silently.

4. Check Domain Skill

Read resources/domain-skill.md if it exists in the project's .claude/skills/ directory. Domain skills define project-specific always-on skills that apply regardless of task type.

5. Load Selected Skills

Invoke each YES skill via the Skill tool immediately. These are now part of your working context.

6. Apply Fallback Chain

If no skill matches what you need:

  1. Improvise — use your general knowledge to handle the task directly
  2. Escalate — if you cannot improvise effectively, consult the operator before proceeding

Do not stall because a skill is missing. The chain is: find skill -> improvise -> escalate.

Output

Produce a Skills to Load list that your caller can reference:

Skills to Load:
- ruby-rails (Rails patterns and conventions)
- pg-perf-power-user (query optimization guidance)

No matching skill found for: Kafka consumer patterns (improvising from general knowledge)

Include both loaded skills and any gaps where the fallback chain was applied.

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.

Check your setup, free