Find repository functions with natural-language search

Parse Python functions, embed an index, and retrieve relevant implementations and paths from a natural-language query.

Goal

Parse Python functions, embed an index, and retrieve relevant implementations and paths from a natural-language query.

Extract function text and paths, then create text-embedding-3-small vectors. [source-openai-find-code-functions-by-natural-language]
Embed a query and return similar function snippets, paths, and nearby context. [source-openai-find-code-functions-by-natural-language]
Test with known functions; use results for navigation, not as proof of correctness or safety. [source-openai-find-code-functions-by-natural-language]

Use known queries to check retrieved function names and paths, then open the source to confirm the implementation rather than a same-named test.

If results are empty or stale, rebuild the index and check whether the parser skipped nested or async functions; show paths without editing code.

Python, OpenAI embedding API, authorized read-only codebase; exclude secret files from the index.

Reads source code from the selected repository and may send function snippets to an embedding API; writes a local index and does not modify repository files.

Run the source on isolated test data, then check outputs and dependency versions before adapting it.

These values describe the upstream repository, not popularity or likes for an individual skill or playbook.

No metrics snapshot is available yet.

License: MIT