Local reviewer-vs-challenger loop that reproduces the GitHub Copilot PR review dynamic before you open the PR. Invokes reviewer-github (a different model) to emulate GitHub's findings, then disputes or resolves each one, giving a per-finding verdict — Fix / Push back / Needs judgment. Read-only.
Emulates GitHub Copilot code review locally. Runs on a different model from the challenger to reproduce the genuine divergence you get from GitHub's own reviewer. Loads the SAME instruction files GitHub reads and produces PR-style findings. Read-only. Invoked by the review agent.
Instructions for microsoft/Agents-for-net, covering copilot instructions for agents-for-net, project overview, build and test, build entire solution and build via top-level project (includes all libraries).
Instructions for microsoft/Agents-for-net, covering code review rules (microsoft 365 agents sdk for .net), what not to comment on, review lenses and severity.
Instructions for microsoft/Agents-for-net: See .github/copilot-instructions.md for build commands, architecture, conventions, and coding guidelines for this repository.
177 5d agoA39 tokens
originalMIT
At most 3 mods per repository are shown here — the rest are on their repository pages: