Infrastructure for models, agents, embeddings, evals, prompts, and AI application backends.
Use these checks to narrow the directory before reviewing individual source files, permissions, and installation instructions.
Check which model providers, embedding formats, observability systems, and deployment boundaries the resource supports. Favor focused components with documented failure handling over broad stacks with unclear dependencies.
Confirm provider APIs, model identifiers, vector dimensions, runtime requirements, and whether the resource assumes a local, cloud, or hybrid deployment.
Common verified inventory signals: Runtimes: Python. Install methods: git clone, npx, and uvx/pip.
Protect model and provider credentials, separate tenant data, review telemetry destinations, and test how retrieved or generated content crosses trust boundaries.
A starting set selected from profiles that pass SkillIndex’s current public index-eligibility checks. Review each profile’s source and setup details before use.
