Agents are principals too
Permissions are a module and action statement resolved over platform, organization and repository scopes. People, agent keys and share links are all principals in the same model, so one model answers who can do what, and where.
Delegation that
cannot widen
An agent key carries its own scopes, intersected with its owner's permissions and re-resolved on every call and every repository. Handing work to an agent can only narrow what is reachable, not broaden it.
An agent's authority is its own scopes intersected with its owner's permissions, resolved per call and per repository.
Included
Related
Security
Advisories assessed for reachability in your code, malware alerts within minutes, remediation pull requests and an AI bill of materials per repository.
Read moreCoding agents
An MCP server any coding agent can call for a review before the pull request exists, and visual plans a person signs off before the agent writes code.
Read moreFleet
One command enrolls a Linux, macOS or Windows machine. inline maps what can reach each port and proposes firewall plans that roll back unless confirmed.
Read more
Put it on one repository
Connect GitHub or GitLab, enable a repo, and the next pull request opened gets a review.