dotrepo MCP server

dotrepo

Repository facts your agents can reuse.

dotrepo provides structured project metadata, build and test commands, documentation links, and ownership information through an HTTP API, MCP server, and CLI. Records carry provenance, evidence pointers, conflicts, and record age so tools can decide what to use and when to inspect upstream sources.

The immediate use is repository orientation and operational metadata. Architecture research, suitability decisions, debugging, and code changes still require source inspection. Making that basic orientation reusable is the first step toward the longer-term mission of reusable repository understanding.

Try a lookup

No installation or maintainer adoption is required for an indexed repository:

curl -sS https://dotrepo.org/v0/repos/github.com/BurntSushi/ripgrep/profile.json

The profile groups purpose, execution, documentation, ownership, completeness, and trust. Check record.generatedAt and record.freshnessStatus before using a value. The top-level freshness.generatedAt dates the export, not the upstream facts. Refreshed records also expose fieldEvidence: source, extraction method, assessment, reason, and check time where the crawler retained them. Older records do not acquire field-level evidence simply by being re-exported.

With dotrepo-mcp installed, call:

{
  "name": "dotrepo.lookup",
  "arguments": {
    "repositoryUrl": "https://github.com/BurntSushi/ripgrep",
    "path": "repo.test"
  }
}

Use dotrepo as a first lookup with explicit fallback. A found profile can still be stale, incomplete, or conflicting. See the integration guide and the runnable consumer example for task-specific checks.

What is proven today

The reference toolchain and hosted lookup surface are implemented. Coverage and quality are still partial. The live measurement page reports field presence and modeled requests; presence is not accuracy, and modeled requests are not measured end-to-end agent savings.

The independent head-to-head benchmark retains both wins and losses. Fixing the examples it exposed is regression evidence, not proof of generalization. A native maintainer record, record-wide verification, and field-level correctness are different properties.

Sustained independent consumer adoption remains an open success criterion. In-repository clients and AI interviews demonstrate or explore an integration; they do not establish external demand.

For maintainers

A native .repo gives you a structured source of truth and tools to keep supported documentation blocks consistent. You can start from existing files or a generated overlay, inspect the result, and publish your own authoritative record. Overlays make indexed repositories usable before adoption; native records provide maintainer control without being required for coverage.

Quick start

or with cargo install dotrepo-cli (pin a 1.0.x version for production). The main branch tracks a 2.0.0-alpha development line with public API changes. See docs/install.md for platform bundles and the VS Code extension package.

1. Install dotrepo from the latest stable GitHub release bundle (1.0.x),

1. Start a record in your repository:

dotrepo --root <repo> init
# or bootstrap from existing README.md / CODEOWNERS / SECURITY.md:
dotrepo --root <repo> import

1. Run the canonical maintainer loop:

Shortened. The full README is on GitHub.