Connect GitHub to Claude Cowork
Give Claude Cowork governed access to GitHub’s 100 tools through Runlayer, with identity, policy, scanning, and audit on every call.
Set up
Add GitHub to Claude Cowork in four steps
Admins approve the server once. Everyone in Claude Cowork then gets it through the same governed connection.
- 1
Approve GitHub in Runlayer
Add GitHub from the catalog and choose who can use it and which of its tools are allowed.
- 2
Connect Claude Cowork to Runlayer
Claude Cowork gets connector tools through the Runlayer Plugin. With direct tool listing turned on for your deployment, GitHub’s 100 tools show up in Claude Cowork’s own tool list.
Claude Cowork setup guide - 3
Sign in to GitHub
Each user signs in to GitHub with OAuth. Runlayer keeps the tokens out of the client’s configuration.
- 4
Use GitHub from Claude Cowork
GitHub’s approved tools appear in Claude Cowork. Every call is checked against policy and recorded.
Tools
100 GitHub tools in Claude Cowork
Runlayer policy decides which of these Claude Cowork users can call.
Get actions
Get details about specific GitHub Actions resources.
List actions
Tools for listing GitHub Actions resources.
Actions run trigger
Trigger GitHub Actions workflow operations, including running, re-running, cancelling workflow runs, and deleting workflow run logs.
Add comment to pending review
Add review comment to the requester's latest pending pull request review.
Add issue comment
Add a comment and/or reaction to a specific issue or issue comment in a GitHub repository.
Add reply to pull request comment
Add a reply and/or reaction to an existing pull request comment.
Assign copilot to issue
Assign Copilot to a specific issue in a GitHub repository.
Assign copilot to issue with intent
Assign Copilot to a specific issue in a GitHub repository.
Control
What Runlayer can see and stop in Claude Cowork
Some controls apply to every client at the gateway. Others depend on what Claude Cowork exposes to Runlayer.
Gateway policy (available)
Every GitHub call from Claude Cowork passes through Runlayer, where your policies can allow, alert, mask, or block it.
Audit trail (available)
Each call is recorded with the user, Claude Cowork as the client, the tool, and the policy and scanner decisions.
Anthropic inference hooks (available)
Anthropic Inference hooks let Runlayer check Claude Cowork prompts before inference.
OTLP telemetry (available)
Claude Cowork can send supported OpenTelemetry events to Runlayer for observe-only session visibility.
Stop a call before it runs (available)
Runlayer can reject a Claude Cowork prompt before Anthropic runs it, which stops the whole request rather than one tool call.
Block output after it runs (available)
Because Runlayer checks the full prompt, tool results in it can be blocked before the model sees them.
Redact output in place (not available)
Runlayer cannot mask content in Claude Cowork. The gateway still applies policy to the call.
In Claude Cowork
- Cowork session data sent over OpenTelemetry is for monitoring only and cannot stop anything.
Frequently asked questions
Connect more