mainline

What problem does it solve? Code history alone does not capture why AI agents made changes, so teams lose the reasoning, abandoned approaches, and constraints behind agent-written code. This Skill records agent intents, decisions, and trade-offs into Git alongside commits, and surfaces semantic conflicts between concurrent agent work before PR review. ## Core Features & Use Cases - Intent lifecycle management: Start, append, seal, and publish intent records that link goals, decisions, and rejected alternatives to the commits that carried the work. - Preflight conflict detection: Run a readiness gate that reports file overlaps, branch drift, and in-flight proposals from other agents before editing code. - Autonomy stop-lines: Enforce team-defined boundaries (assist, handoff, review) so agents stop before commits, pushes, PRs, merges, or deploys unless explicitly authorized. - PR body generation: Generate pull request descriptions from sealed intents so reviewers see the original goal and reasoning, not just the diff. - Use Case: Two agents are modifying the same billing logic on different branches. Before editing, each runs mainline preflight --json, sees the other's in-flight intent, and narrows scope or syncs instead of producing a semantic conflict Git cannot detect. ## Quick Start Ask the agent to run mainline preflight, start an intent for your task, and seal the work with a summary once the commit is ready.