What problem does it solve? When an AI agent implements a feature autonomously, it can lose progress on interruption, diverge from the agreed plan, skip tests, or break downstream automation by creating pull requests incorrectly. This Skill operationalizes the Implement stage of the Fabrik SDLC pipeline so code changes are committed incrementally, tested, pushed, and handed off to the engine in a predictable format. ## Core Features & Use Cases - Plan-driven execution: Reads the task checklist from .fabrik-context/stage-Plan.md and works through tasks in order, resuming from prior commits instead of restarting. - Disciplined Git workflow: Enforces one Conventional Commit per task, frequent pushes, and task check-off in the Plan stage comment via the GitHub CLI. - Engine-coordinated PR creation: Signals pull request creation through a FABRIK_PR_CREATE_BEGIN/END marker instead of calling gh pr create, so the engine owns the Closes #N line. - Safe long-running verification: Prescribes foreground, timeout-bounded test and dev-server patterns so no orphaned processes outlive the stage. - Use Case: An issue planned by the Fabrik Plan stage is picked up; the agent implements each task with tests, commits and pushes incrementally, updates the checklist comment, emits the PR marker, and finishes with FABRIK_STAGE_COMPLETE. ## Quick Start Invoke this skill as the Fabrik Implement stage agent to execute the plan in .fabrik-context and produce committed, tested, pushed code on the issue branch.
