GitHub
MCP server github/github-mcp-server/io.github.github/github-mcp-server
Connect AI assistants to GitHub - manage repos, issues, PRs, and workflows through natural language.
No install line here. This server does not publish one we could copy, and it has no package name recorded, so anything shown here would be a guess at whether it runs from npm, PyPI, a container or a hosted URL.
The repository will have the real instructions: github.com/github/github-mcp-server
In 4 stacks
Whose pick this was
Write against the API that exists today
The second opinion, and it is a different clock rather than a backup. Documentation is a rendering of the code and it lags. Release notes, tags, and a closed issue titled "removed in v5" are the code. Reach for this the moment the docs are silent on something, because silence in documentation is not evidence of absence.
agentscliIn house
Read a codebase you did not write
The "why is it like that" tool, and the one people leave out. A weird function is usually a fixed bug. The pull request that introduced it says so, and without history the agent reads the workaround as a mistake and helpfully removes it.
agentscliIn house
Find the cause before writing the fix
Then find out why the code is shaped like that, which is the step that stops the fix from being a regression. The odd conditional you are about to simplify is usually a scar. The pull request that introduced it says so in its title, and that sentence exists nowhere in the working tree.
agentscliIn house flow
A review that is not a rubber stamp
Start by getting the things a diff does not contain. Whether the issue in "fixes #482" describes what the patch does. Whether the checks ran at all, or ran red. Whether the function whose signature just changed has three other callers elsewhere in the repository. All four of those are review findings and none of them are visible in the patch text.
agentscliIn house flow