What problem does it solve? Coordinating multiple AI agents on one shared change often leads to file collisions, unverified claims of completion, and handoffs that omit critical evidence. This Skill structures multi-agent work into explicit lanes with ownership rules, status vocabulary, and gates so parallel work stays safe and auditable. ## Core Features & Use Cases - Lane orchestration with WIP collision detection: Define a Team Contract and lane map, detect dirty-file conflicts via git status before any builder opens a file, and sequence or merge colliding lanes. - Quality gates and escalation thresholds: Run planning, review, and release gates as advisory recommendations, with hard stop conditions for repeated fix cycles, security findings, cost spikes, and contradictory constraints. - Public contribution program: Score, lease, and track repository contribution tasks through a deterministic JSON ledger script with atomic locking, one-shot publication grants, and outward artifact checks. - Use Case: A team needs to ship an auth rewrite safely. The orchestrator requires an accepted RFC, assigns a builder to auth files only, runs code grading and review lanes, verifies the release in production, and closes with a signed evidence handoff. ## Quick Start Use suede-agent-teams to split my feature work into coordinated lanes with a team contract, collision detection, and a final evidence handoff.
