cocoindex-code vs codebase-memory-mcp
AST-based semantic code search for coding agents: pipx install, zero config, local embeddings out of the box — a CLI/skill/MCP that cuts agent context ~70% vs grepping. Built on CocoIndex. — versus — Code intelligence MCP in pure C: tree-sitter knowledge graph over 158 languages, average repo indexed in milliseconds, sub-ms queries, 10x fewer tokens. Single static binary, zero deps.
The two query styles of agent code context: cocoindex-code embeds AST chunks for semantic 'find the code that does X'; codebase-memory graphs symbols for structural 'who calls this and what breaks'. Different questions — many stacks want both.
| cocoindex-code | codebase-memory-mcp | |
|---|---|---|
| Stars | 2.6k | 34k |
| Forks | 203 | 2.6k |
| Language | Python | C |
| License | Apache-2.0 | MIT |
| Last activity | 5 days ago | yesterday |
| Topics | coding, local | coding, local |
| Curated connections | 4 | 3 |
cocoindex-code — the curator's take
The lowest-friction entry in the code-context-server category: one pipx install, zero config, and the [full] variant ships local embeddings so there's no API key before first search. AST-aware chunking plus the CocoIndex engine underneath means the index stays fresh incrementally instead of rebuilding. Integrates as a skill or MCP with Claude Code, Codex, Cursor. NOT the deepest graph: it's semantic search, not call-graph analysis — no callers/blast-radius queries (that's TokenSave or codegraph-mcp territory) — and the slim install quietly requires a cloud embedding key while [full] costs ~1GB of torch.
codebase-memory-mcp — the curator's take
The performance ceiling of the code-context-server category: pure C, single static binary, the Linux kernel indexed in 3 minutes, structural queries under a millisecond — with a peer-reviewed preprint (83% answer quality, 10x fewer tokens across 31 repos) instead of vibes. Hybrid LSP adds real type resolution for the 12 languages that matter most, and 43 client surfaces means it plugs into whatever agent you run. NOT semantic search — it answers structural questions (call chains, routes, blast radius), not 'where's the code that does X'; pair it with an embedding tool for that. And note its own disclosure: it writes to your agent config files by design — audit posture is unusually good (SLSA 3, OpenSSF, per-release VirusTotal), use it.