Skip to main content
Glama
AlbertoMitroi

devos-intelligence-mcp

devos-intelligence-mcp

An MCP server that answers AI agents (Claude Code, Codex) from the intelligence generations of the repository they work in: semantic graph, observations and context packs written by devos-dotnet-intelligence and devos-angular-intelligence in the shared intelligence contract. The agent never receives a generation, only compact answers with file:line evidence and short redacted snippets.

Status: 14 tools over index.sqlite and context packs; generation resolved from local folders, the OCI cache or GHCR (context-<repo>:<sha>-<engine>, private packages through the environment, the Docker credential store or gh), nearest analyzed ancestor announced in every answer; the index is read by contract v1.2.0 index.sql (vendored, drift-checked), the optional evidence tables and site are used when present. Federation system generations answer across repositories (who_calls, trace, messages), each location labelled with its repository. Tested against the real public packages on ghcr.io for devos-lab-web (Angular) and devos-lab-leasing (.NET). Unit and integration tests pass locally (scripts/verify.sh), including the official MCP SDK client and a registry:2 round trip in the release layout. Public at AlbertoMitroi/devos-intelligence-mcp; CI runs the unit tests, the image and anonymous pulls from ghcr.io on amd64 and arm64.

Use

docker build --target runtime -t devos-intelligence-mcp:local .
claude mcp add devos-intelligence --scope user -- ~/repos/devos/devos-intelligence-mcp/scripts/mcp-docker.sh

Codex, configuration and tool list: docs/USAGE.md.

Related MCP server: engram-serve

Develop

Docker and git are needed; the tests read the sibling folders (_generations, devos-lab-web).

scripts/verify.sh

With Node.js 24.22 or later on the host, npm ci then npm test runs the unit tests.

Documents

License

MIT, see LICENSE.

Related MCP Connectors

Related MCP Servers

  • A
    license
    Not graded
    quality
    B
    maintenance
    Enables coding agents to query a local, versioned knowledge graph of a software project, retrieving overviews, context packs, evidence, and explanations to make informed changes.
    MIT
  • A
    license
    Not graded
    quality
    A
    maintenance
    Enables AI coding agents and IDEs to query a local structural code graph and retrieve focused, token-bounded context on code entities, Git relationships, callers/callees, impact, paths, and recorded mistakes for more efficient sessions.
    35 npm
    MIT
  • A
    license
    Not graded
    quality
    A
    maintenance
    Enables AI coding agents to retrieve bounded, focused context from a local structural code graph, including source entities, Git relationships, and recorded project memory. It supports exploring structure, inspecting callers, callees, and impact, and carrying prior mistakes into later sessions without reading full files.
    35 npm
    MIT
  • A
    license
    Not graded
    quality
    A
    maintenance
    Enables AI coding agents to query a local structural code graph and receive focused, token-bounded context about source entities, relationships, Git co-change, and selected project memory without reading full files.
    35 npm
    MIT