Works best for larger code bases - from 500 files upwards.
Use it as MCP with e.g. Claude Code or Cursor, super fast and includes tracing for agentic systems (finding callers and callees directly so that agents know faster where to look at).
I tried to include much of the documentation, use as cli and --help to get more insight! :)
Currently Python only; JS/TS and others are on the roadmap. Soon to be published as python package (pypi).
Would love feedback from folks with large monorepos or heavy AI-agent workflows!
Semantic fingerprints for 100x faster Python code search.
Symdex-100 indexes every function in your repo into a small SQLite sidecar (`.symdex/index.db`). Each function gets a structured 20-byte “Cypher” (e.g. `SEC:VAL_TOKEN--ASY` = security, validates token, async) instead of opaque embeddings. You search by intent—“where do we validate user tokens”—and get sub-second, ranked results from the index. Source files are never modified.
