No-train, on-prem code knowledge graph served to AI agents over MCP, with a hash-chained audit row for every read.
cognis-digital/codegraph-mcp is an MCP server distributed on github, maintained by cognis-digital, tracked here since June 2026. It has shipped 1 release and exposes 11 tools. Tools include cross_language_edges, find_callees, find_callers, find_hotspots, find_orphans, find_references, and 5 more. Its composite risk grade is A — an inferred review prompt computed from observed signals, not a verdict.
Reading the source raised one review prompt — exfiltration combo — each a pattern worth a human look rather than a finding of fault.
Since the previous scan on 2026-08-26: 3 tools added.
full security breakdown →-
cross_language_edgesAll resolved cross-language HTTP edges. -
find_calleesWhat a symbol calls. -
find_callersDirect callers of a symbol — includes cross-language edges. -
find_hotspotsMost depended-on symbols (highest caller count) — where changes ripple furthest. -
find_orphansDead-code candidates: functions/methods with no callers and not HTTP entrypoints. -
find_referencesEvery call site / use of a name. -
get_symbolFull record for one symbol id (signature, location, container). -
graph_statsFile / symbol / edge / language counts. -
impact_analysisTransitive callers — the blast radius of a change. -
project_graphModule/package-level dependency graph — the architecture map above the symbols. -
search_symbolsFind symbols by name substring (optionally filter by kind).
06-14 00:00 07-28 00:00 now
| version | date | src |
|---|---|---|
| walkthrough-v1 | 2026-06-30 | github |