Claude Mates runs AI agents that consume API tokens. The CLAUDE_MATES_API_KEY secret is protected by GitHub's built-in security model:

• Fork PRs never receive secrets — GitHub strips all secrets from fork-originated workflow runs
• Only push: main and workflow_dispatch trigger the workflow — not pull_request
• Branch protection requires review before merge to main — no direct pushes
• CODEOWNERS requires owner approval for workflow, runner, and config changes

Repository owners should:

1. Set a monthly spend limit on the API key via console.anthropic.com
2. Use a dedicated API key for Claude Mates (not shared with production services)
3. Monitor usage via the Anthropic dashboard

• The Claude Code CLI is pinned to a specific version in the workflow
• When used as an external action, the checkout is pinned to a commit SHA
• Branch protection prevents tampering with pinned references

If you discover a security issue, please report it via GitHub Security Advisories (not public issues): https://github.com/vlad-ko/claude-mates/security/advisories/new

• Never merges PRs — all changes require human approval
• Never accesses secrets directly — Claude Code CLI has no Bash(env *) or Bash(echo $*) tools
• Never modifies workflow files — CODEOWNERS blocks this
• Never runs on fork PRs — workflow only triggers on push to main
• Max 1 PR per mate per day — prevents spam even if compromised