agentskills/agentskills

• **Archetype**: methodology-repo
• **Stars**: 15,904
• **Last pushed**: 2026-04-10
• **License**: Apache-2.0
• **Discovered**: 2026-04-12
• **Skills found**: 0 (specification repo, no SKILL.md files)

Summary

The official Agent Skills specification and documentation repository maintained by the agentskills organization. Defines the SKILL.md format, client implementation guidelines, and the overall skills ecosystem standard. Contains specification docs, a Python SDK reference implementation (skills-ref/), and client implementation guides. This is the standards body, not a skills collection.

Assessment

Critical reference for understanding the agent skills specification that all other repos implement. The specification defines the SKILL.md format, metadata schema, and client integration contracts. The Python reference implementation in skills-ref/ provides validation patterns. The client implementation docs explain how harnesses should discover, load, and execute skills -- directly relevant to babysitter's harness adapter layer and skill:discover command.

Extraction Priority

• Medium
• Rationale: No extractable skills or processes, but the specification itself is essential reference material for babysitter's skill discovery and execution infrastructure. The client implementation guide should inform harness adapter improvements.

Processes

• No direct process extraction -- this is a specification, not a skills collection
• The specification's skill lifecycle (discover -> load -> validate -> execute) could inform a methodologies/skill-lifecycle-validation.js process for validating skill packs

Plugin Ideas

• **DevX**: Skill validator plugin that validates SKILL.md files against the official agentskills specification, checking frontmatter schema, required fields, and tool declarations
• **workflow automation**: Skill conformance checker plugin that audits installed skills for spec compliance and reports deviations

Patterns

• YAML frontmatter specification with required fields: name, description; optional: license, allowed-tools, metadata
• Client implementation pattern: discovery via filesystem walk, SKILL.md parsing, tool permission gating
• Python reference implementation using standard packaging (pyproject.toml, uv.lock)
• Mintlify-based documentation (docs.json, .mdx files) for the specification website at agentskills.io

Library Mapping

Plugin Marketplace Mapping