axm-echo
Find similar documented Python symbols across packages, or search for a helper matching an intention. Echo retrieves evidence for reuse and deduplication; it does not prove semantic equivalence or change source files.
| Need | Entry point |
|---|---|
| Find cross-package duplicate candidates | axm echo_code |
| Search before implementing a helper | axm echo_check --intention "..." |
| Embed text or extract a Python corpus | from axm_echo import embed, extract_package |
| Compare normalized Python statement shapes | statement_set and jaccard_similarity |
Start here
Install in a Python 3.12+ project with uv add axm-echo, then follow the
getting-started tutorial. It uses a temporary
corpus and TF-IDF, so it needs no model download.
The tools default to the neural st backend; the Python embed() function
defaults to tfidf. Neural dependencies are part of the base install.
Architecture and limits explains the runtime
costs, model cache and limits of similarity scores.
Find your way
- Tutorial: extract and compare a controlled corpus.
- How-to: configure scope, reuse during planning, review clusters, compare structural shapes.
- Reference: CLI and tools, result contracts, Python API.
- Explanation: pipeline, tradeoffs and limits.
The corpus covers Python functions and classes exposed by their defining modules. Both tools require docstrings; undocumented implementations are not searched. Always check the resolved scope and the actual code before deciding that an empty report permits a new implementation.