Skills Catalog Intake & AEEF Starter Mapping

This reference page shows how to intake an external skills catalog (for example, awesome-claude-skills) and convert selected entries into an AEEF-governed internal starter catalog.

It complements:

• PRD-STD-017: Agent Skills Catalog Governance
• Skills Catalog Enterprise Usage
• Skill Registry & Multi-Agent Gate Patterns

What Was Implemented (Example Intake)

1. Quarantine Clone

• External source cloned to: intelligence/intake/quarantine/awesome-claude-skills
• Purpose: evaluate metadata and candidate skills without direct runtime adoption

2. Attribution Manifest

• Manifest: intelligence/intake/manifests/awesome-claude-skills_attribution_manifest.md
• Includes source owner, URL, commit snapshot, category counts, and attribution obligations

3. Triage Evaluation

• Evaluation record: intelligence/intake/evaluations/awesome-claude-skills_aeef_triage_2026-02-26.md
• Decision classes: APPROVED, PROVISIONAL, BLOCKED_FOR_STARTER, WATCHLIST

4. Internal Starter Catalog Mapping

• Mapped catalog file: reference-implementations/transform/skills-catalog/aeef-approved-starter-catalog.yaml
• AEEF fields included: skill-id, risk class, role/environment gates, required gates, standards mapping, attribution source

Additional Intake Example: Official Vendor Skills (anthropics/skills)

In addition to community discovery catalogs, AEEF also supports intake of official vendor skill repositories.

Implemented example:

• Quarantined clone: intelligence/intake/quarantine/anthropics-skills
• Attribution manifest: intelligence/intake/manifests/anthropics-skills_attribution_manifest.md
• Triage record: intelligence/intake/evaluations/anthropics-skills_aeef_triage_2026-02-26.md
• Per-skill license matrix: intelligence/intake/manifests/anthropics-skills_license_matrix.md

Why this matters:

• gives AEEF an official baseline for skill packaging and quality patterns
• improves evaluation of community skills against vendor examples
• requires license-aware intake because some skills may be source-available rather than open source

Starter Catalog Design Principles (AEEF)

• Start small: 5-10 skills with clear owners and strong traceability
• Approve low-risk workflow skills first (testing, debugging, review support)
• Keep elevated/mutating skills provisional until sandboxed validation is complete
• Preserve upstream attribution for every mapped skill
• Track per-skill license/usage status (especially for official vendor repos with mixed terms)
• Bind skills to role and environment gates before enabling multi-agent execution

Selected Mapping Snapshot (from intake)

AEEF Skill ID	Upstream Skill	Status	Primary Use	Key Controls
aeef.impl.tdd-assist	test-driven-development	Approved	feature implementation discipline	code review + tests
aeef.reliability.root-cause-tracing	root-cause-tracing	Approved	debugging and failure analysis	read-first usage + human review
aeef.reliability.systematic-debugging	systematic-debugging	Approved	structured debugging	test validation + review
aeef.testing.pairwise-test-design	pypict-claude-skill	Approved	risk-based test generation	test gate + review
aeef.review.plan-implementation-conformance	review-implementing	Approved	plan/spec conformance review	no merge authority
aeef.git.worktree-isolation	using-git-worktrees	Provisional	parallel agent work isolation	dev-only, no merge/push
aeef.qa.webapp-testing	webapp-testing	Provisional	browser testing automation	staging-only + sandboxed browser
aeef.security.vibesec-review	VibeSec-Skill	Provisional	secure vibe coding review	security owner review
aeef.security.owasp-review	owasp-security	Provisional	security review companion	sensitive skill gates
aeef.security.secret-safe-session	varlock-claude-skill	Provisional	secret hygiene workflow	secret-safe test harness

How to Apply This Pattern in Your Repo

1. Clone external catalog into intake/quarantine (never directly into runtime skills/ folders).
2. Generate an attribution manifest with source URL, owner, snapshot commit, and parser output.
3. Triage candidate skills against AEEF pillars and PRD standards.
4. Map selected skills into an internal registry file with role/environment gates.
5. Enable only approved skills in developer and agent workflows.
6. Keep provisional skills sandboxed and behind explicit human approval.

Integration with Agent SDLC

Use the mapped catalog together with:

• Agent SDLC Setup for registry/contracts/handoff templates
• OpenClaw for AEEF Agent Orchestration if you want a gateway-based multi-agent execution surface

Attribution Notice

This page does not copy external skill content into AEEF. It demonstrates governance-first intake, attribution, and controlled mapping of selected skills into an internal enterprise catalog.