Tricorder

Turns a repository's code-review history into a map of what a team knows, what it repeats, and where review attention is thin.

PythonGitHubAnthropicGemini

The problem

Review comments hold a team’s real standards, and they disappear into merged pull requests. Nobody can see which corrections keep recurring or who the expertise sits with.

What it does

Tricorder reads local repository evidence, git history and review discussions, and produces an inspectable map of recurring standards, expertise concentration, gaps, and places where knowledge could move into education, tooling or automation. It runs in six levels, and the first three work without any API key.

Three published runs are in the explorer: cal-itp/data-infra (190 pull requests), block/buzz (272) and block/berd (163). On block/berd, every pull request that touched the Tauri capabilities manifest or the CSP configuration, measured on 2026-09-03, merged with no human comment on those files.

Impact Advocate (beta)

A companion agent skill that turns your own GitHub history into an evidence-linked self-review draft. It has its own page: Impact Advocate (beta).

Caveats

The model-generated judgments are experimental and need human review. Tricorder is not a performance-review or developer-ranking tool.

Explore the findings · Source