What problem does it solve?
When two AI agents work in parallel on the same repository, their branches inevitably collide. Each agent lacks the peer's context and tends to either overwrite the peer's work or abandon its own changes. This Skill provides an impartial reconciler that reads both diffs and both sides' stated intents, then produces a merged result without favoring either party.
Core Features & Use Cases
- Hunk Classification: Categorizes every conflict marker as disjoint-intent, same-question-different-answer, or superseded, then applies the correct resolution rule.
- Impartiality Contract: Operates as a third-profile spawned agent so the reconciler is structurally unbiased toward either original worker.
- Intent-Aware Decisions: Pulls completion summaries from kanban cards or PR bodies to judge design questions, never splitting the difference into a hybrid nobody designed.
- Use Case: Two agents in a parallel-PR wave both modify the same authentication module. Spawn this arbiter with both branch names and both kanban summaries to produce a single merged commit with every design decision explicitly documented.
Quick Start
Spawn a neutral subagent with the repo path, both branch names, and both sides' intent summaries, instructing it to follow the agent-merge-conflict-arbiter skill.
