What problem does it solve? It turns an OpenSpec change's task list into actual code changes by guiding the AI through selecting the change, reading its context files, implementing each task, and tracking progress until completion or a blocker. ## Core Features & Use Cases - Change Selection: Infers the target change from context, auto-selects when only one is active, or lists available changes for the user to pick. - Status-Driven Implementation: Reads schema, progress, and apply instructions from the openspec CLI, then works through pending tasks one by one, marking each - [x] as completed. - Guardrails and Pause Handling: Pauses on ambiguity, design issues, scope creep, or errors instead of guessing, and reports progress with structured output. - Use Case: After drafting a spec-driven change with proposal, design, and tasks, ask the agent to implement it; it reads the artifacts, writes the code, and checks off tasks until all are done. ## Quick Start Ask the agent to implement the OpenSpec change named add-auth and work through its tasks.