external-comms

To draft GitHub issue responses for human review, this Skill scans open issues and discussions within a default seven-day window, classifies them, and generates template-based replies while excluding internal or wontfix items. It then presents drafts in a standardized review table requiring explicit human approval before posting.

Yes, you can automate GitHub community triage with audit logging. This Skill scans unanswered open items, drafts template-based replies, and logs every action to the .squad/comms/audit/ directory, ensuring all automated triage steps are recorded for rollback and compliance before any human approval gate.

This community response workflow uses MCP tools to scan open GitHub issues and discussions, with the gh CLI serving as a fallback mechanism. Both methods identify unanswered items within seven days, exclude internal or wontfix labels, and verify thread reads before drafting template-based responses.

The best way to manage human-reviewed GitHub replies is using an explicit approval gate. This Skill prepares CLI commands for posting but requires a human operator to issue explicit approval commands before any action occurs, preventing automated posting of drafts generated from T1–T9 templates.

Thread-read verification requires confirming the full context of an open GitHub issue or discussion before generating a draft response. This Skill enforces this check to ensure template-based replies address the actual conversation history, flagging long threads and sensitive signals for human escalation during the review process.

Rollback controls for bad GitHub posts include deletion and replacement procedures managed through the audit system. Every drafted and posted action is logged to .squad/comms/audit/, allowing operators to track, flag, and reverse sensitive or incorrect community responses after they pass the human approval gate.