[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"readme:codegraph-mcp":3},"\u003Ch1>codegraph-mcp\u003C\u002Fh1>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fcognis-digital\u002Fcodegraph-mcp\u002Factions\u002Fworkflows\u002Fci.yml\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fgithub.com\u002Fcognis-digital\u002Fcodegraph-mcp\u002Factions\u002Fworkflows\u002Fci.yml\u002Fbadge.svg\" alt=\"CI\" \u002F>\u003C\u002Fa>\u003C\u002Fp>\n\u003Cblockquote>\n\u003Cp>Part of the \u003Cstrong>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fcognis-digital\u002Faccountable-ai-suite\" rel=\"nofollow ugc noopener\">Accountable AI Engineering suite\u003C\u002Fa>\u003C\u002Fstrong> — provable governance for AI agents on infrastructure you own.\u003C\u002Fp>\n\u003C\u002Fblockquote>\n\u003Cp>\u003Cstrong>A no-train, on-prem code knowledge graph that you serve to AI agents over MCP — with a hash-chained audit row for every read.\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Cp>Ask yourself:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Do your AI coding agents need to understand a codebase that \u003Cstrong>can't be uploaded to a vendor's cloud\u003C\u002Fstrong>?\u003C\u002Fli>\n\u003Cli>Have you watched an agent miss that a one-line change breaks a caller \u003Cstrong>in another language\u003C\u002Fstrong> — because it only ever saw one file?\u003C\u002Fli>\n\u003Cli>When an agent reads your source, could you produce \u003Cstrong>a record of exactly what it read\u003C\u002Fstrong> — and prove it wasn't edited after the fact?\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>If that's your world, you're in the right place. \u003Ccode>codegraph-mcp\u003C\u002Fcode> gives agents real, \u003Cem>structural\u003C\u002Fem> understanding of your code, on hardware you control:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>No training, ever.\u003C\u002Fstrong> The graph exists only to answer queries. Your code is never used to rank, sell, or train a model, and nothing leaves the machine.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Overlay, not migration.\u003C\u002Fstrong> Point it at any checkout or git URL. Keep hosting your code exactly where it already lives — GitHub, GitLab, an internal mirror, an air-gapped drive.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Provable reads.\u003C\u002Fstrong> Every query an agent makes lands in a tamper-evident, hash-chained audit log you can verify offline. \"Which agent read what, and when\" is a fact you can show a regulator, not a guess.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Sees what one-file context can't.\u003C\u002Fstrong> Six languages, with cross-language edges — the dependency a giant context window misses. (Independent research keeps finding graph-structured understanding beats stuffing raw files into a prompt.)\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Chr \u002F>\n\u003Ch2>Watch the walkthrough\u003C\u002Fh2>\n\u003Cp>A full tour — browsing the repo, the setup, the live code graph, and all five demo scenarios running for real (narrated, ~12 min):\u003C\u002Fp>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fcognis-digital\u002Fcodegraph-mcp\u002Freleases\u002Fdownload\u002Fwalkthrough-v1\u002Fwalkthrough.mp4\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fraw.githubusercontent.com\u002Fcognis-digital\u002Fcodegraph-mcp\u002FHEAD\u002Fmedia\u002Fwalkthrough-thumb.png\" alt=\"Watch the codegraph-mcp walkthrough\" \u002F>\u003C\u002Fa>\u003C\u002Fp>\n\u003Cp>▶ \u003Cstrong>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002Fcognis-digital\u002Fcodegraph-mcp\u002Freleases\u002Fdownload\u002Fwalkthrough-v1\u002Fwalkthrough.mp4\" rel=\"nofollow ugc noopener\">Watch the walkthrough (MP4)\u003C\u002Fa>\u003C\u002Fstrong> — click the thumbnail or the link to play.\u003C\u002Fp>\n\u003Ch2>What it does\u003C\u002Fh2>\n\u003Cp>On every push or re-index, \u003Ccode>codegraph-mcp\u003C\u002Fcode> parses your source and builds a queryable graph of:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>Symbols\u003C\u002Fstrong> — functions, methods, classes, and types, with signatures and exact locations.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Call edges\u003C\u002Fstrong> — who calls whom, within and across files.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Cross-language edges\u003C\u002Fstrong> — the edge nobody else resolves: a TypeScript \u003Ccode>fetch('\u002Fapi\u002Fusers\u002F:id')\u003C\u002Fcode> linked to the Go \u003Cem>and\u003C\u002Fem> Python handlers that serve that route. These files share no symbol name, so only a structural join finds the dependency.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>References\u003C\u002Fstrong> — every call site \u002F use of a name.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>Then it answers the questions an agent (or a human) actually asks: \u003Cem>find this symbol, who calls it, what's the blast radius if I change it, what crosses a language boundary here.\u003C\u002Fem>\u003C\u002Fp>\n\u003Ch2>Quick start\u003C\u002Fh2>\n\u003Cpre>\u003Ccode class=\"language-bash\">git clone https:\u002F\u002Fgithub.com\u002Fcognis-digital\u002Fcodegraph-mcp\ncd codegraph-mcp\npip install -e .          # or just run via `python -m codegraph`\n\n# 1. Index any repo (a local path, or a git URL it clones read-only)\ncodegraph index .\u002Fexamples\u002Fsample_repo --db graph.db\ncodegraph index https:\u002F\u002Fgithub.com\u002Fyour-org\u002Fyour-service.git --db graph.db\ncodegraph index . --since HEAD~1 --db graph.db   # incremental: only what changed\n\n# 2. Query the graph\ncodegraph query search loadUser --db graph.db\ncodegraph query impact 7 --db graph.db          # transitive callers (\"blast radius\")\ncodegraph query xlang --db graph.db             # cross-language HTTP edges\n\n# 2b. Visualize the graph — Mermaid (renders inline on GitHub) or Graphviz DOT\ncodegraph viz --db graph.db --view project --format mermaid\ncodegraph viz --db graph.db --view impact --symbol 7 --format me\n\u003C\u002Fcode>\u003C\u002Fpre>\n",1784240407003]