symdey-100 – Intent-based Python code search with semantic fingerprints

Summary: symdey-100 indexes every Python function in a repository into a compact SQLite sidecar using structured 20-byte semantic fingerprints called Cyphers. It enables sub-second, intent-driven code search by querying function semantics rather than keywords, improving search speed and reducing token usage for AI agents.

What it does

symdey-100 extracts per-function metadata from Python ASTs and assigns Cyphers via an LLM or fallback rules. It stores these in a SQLite index that supports multi-lane retrieval and call graph queries, allowing precise, ranked intent-based searches without modifying source files.

Who it's for

Developers and AI agents needing fast, semantic code search within Python repositories to locate functions by intent rather than keywords.

Why it matters

It addresses the inefficiency of grep and full-text search by compressing function semantics into a queryable index, enabling up to 100x faster lookups and 50x fewer tokens consumed during AI code exploration.