agentic-awesome-skills

How does a coding agent pick skills without the tool ranking them? AAS Core lets the agent own the choice, then validates the selection from a local catalog of over 2,000 skills.

The description on GitHub is dense but specific: AAS Core is a local, agent first control plane for complete catalog discovery, agent owned selection, stack validation, and planning, backed by 2,005 plus agentic skills. The README calls it a set of local, agent owned skill stacks for coding agents, from complete catalog access to a reproducible, reviewable plan. The project is Python under the MIT license.

The README states the current release is V15.13.0, which includes AAS Core for local catalog search, agent owned selection, manifest validation, planning, and diagnosis. Apply and recovery are described as experimental and outside the supported preview path. Version numbering like this suggests a project moving quickly and being careful about what it promises as stable.

The selection flow is the heart of the design. A coding agent inspects the project and chooses exact skills from the complete local catalog. AAS Core does not rank or recommend skills. Its read only compose stack tool validates the agent owned selection in memory, and a client or the CLI can persist it as a stack file and produce an immutable plan before any target change. The refusal to rank skills is a deliberate stance: the agent owns the choice, the tool just validates it.

The boundaries are drawn explicitly. This is an independent community project not affiliated with or endorsed by Google, and product names are referenced only to describe compatibility. The GitHub repository is canonical, while the hosted catalog and browser local Workbench are companion discovery surfaces rather than a hosted control plane. That framing keeps expectations honest about what runs where.

• Source 1 · Project repository
• Source 2 · README