codebase-memory-mcp and srclight

These are competitors offering different trade-offs: codebase-memory-mcp prioritizes lightweight indexing efficiency (sub-ms queries, single binary, 99% token reduction) while srclight offers richer analysis capabilities (call graphs, git blame, build system analysis, semantic search) at the cost of higher complexity.

codebase-memory-mcp
63
Established
srclight
51
Established
Maintenance 25/25
Adoption 10/25
Maturity 9/25
Community 19/25
Maintenance 13/25
Adoption 5/25
Maturity 18/25
Community 15/25
Stars: 585
Forks: 65
Downloads: —
Commits (30d): 317
Language: C
License: MIT
Stars: 12
Forks: 5
Downloads: —
Commits (30d): 0
Language: Python
License: MIT
No Package No Dependents
No risk flags

About codebase-memory-mcp

DeusData/codebase-memory-mcp

MCP server that indexes your codebase into a persistent knowledge graph. 64 languages, sub-ms queries, 99% fewer tokens than grep. Single Go binary, no Docker, no API keys.

Builds an AST-based knowledge graph using tree-sitter parsers with optional LSP-style type resolution for Go, C, and C++, persisting the graph to in-memory SQLite for sub-millisecond structural queries. Indexes codebases at extreme speed through RAM-first pipeline with LZ4 compression and fused Aho-Corasick pattern matching, completing the Linux kernel in 3 minutes. Implements the Model Context Protocol with 14 tools including architecture analysis, call graph tracing, impact mapping from git diffs, and Cypher-like graph queries—integrating with 10 coding agents (Claude Code, Zed, Gemini CLI, and others) through automatic MCP configuration on install.

About srclight

srclight/srclight

Deep code indexing MCP server for AI agents. 25 tools: hybrid FTS5 + embedding search, call graphs, git blame/hotspots, build system analysis. Multi-repo workspaces, GPU-accelerated semantic search, 10 languages via tree-sitter. Fully local, zero cloud dependencies.

Implements a filesystem-watched SQLite + tree-sitter indexing layer that increments on git changes via post-commit hooks, with optional Ollama embeddings cached as `.npy` matrices for GPU-accelerated cosine similarity in ~3ms. Operates as a stdio or SSE MCP transport server compatible with Claude Code and Cursor, supporting multi-repo workspaces through SQLite ATTACH and RRF-ranked hybrid search across keyword (FTS5/trigram) and semantic results.

Related comparisons

codebase-memory-mcp and octocode-mcp codebase-memory-mcp and mcp-codebase-index codebase-memory-mcp and code-memory codebase-memory-mcp and deep-code-reasoning-mcp codebase-memory-mcp and heuristic-mcp codebase-memory-mcp and CodeMCP

Scores updated daily from GitHub, PyPI, and npm data. How scores work