What problem does it solve? When a per-repo dev-team session crashes mid-sprint, the orchestrating CTO session loses its worker and the sprint stalls. This Skill recovers the crashed session by reading the recorded session ID and relaunching it with claude --resume, avoiding a full restart and lost context. ## Core Features & Use Cases - Session Recovery: Reads the repo's .claude/current-session.id, verifies the matching JSONL transcript exists, and opens a new Terminal tab running claude --resume <session-id>. - Fleet-Aware Anchoring: Resolves the CTO home via multiple validated anchors (environment variables, .cto-path files, ~/.cto/home) and looks up the repo path in the .cto/projects.yaml registry. - State Cleanup: Clears the repo's CRASHED flag and re-applies the pinned model so the resumed session matches the original sprint configuration. - Use Case: /sprint-status reports repo api-service as CRASHED. Run this Skill with the repo name to relaunch its session in a new Terminal tab, then verify progress with /peek. ## Quick Start Ask the agent to resume the crashed dev-team session for a specific repo by name, for example: resume the dev team session for repo api-service.
