Human-in-the-loop tooling for AI coding agents — pause for review, approve risky actions, and steer long runs.
HumanLayer adds a human review layer to AI coding agents working on hard problems in complex codebases. Features: - Approval gates for risky tool calls (deletions, deploys, billing) - Async human review while the agent waits - Slack / web / email approval channels - Audit trail of what was approved by whom - Long-running session support — agents pause, resume after human input - Works with Claude Code, OpenAI agents, custom agent frameworks Use it when fully autonomous is too risky but you don't want to babysit every action. 10,500+ GitHub stars.
Install this plugin from the source repository:
git clone or copy-paste into your ~/.claude/skills/ directory).No comments yet. Be the first to share your thoughts!
Build, debug, and optimize apps with the Claude API and Anthropic SDKs for Python and TypeScript.
AI code review agent for GitHub and GitLab PRs — understands full codebase context.
Professional security skills: code auditing, vulnerability detection, smart contract security, and reverse engineering.