speckit-implement_skill

Overview

This skill executes an implementation plan by processing and running every task defined in specs/<feature>/tasks.md and the project plan. It automates prerequirement checks, checklist validation, task orchestration, and updates task files as work completes. The skill produces code changes, updated ignore files, and a final status summary tied to the original specification.

How this skill works

It starts by running the repository prerequisite script to discover FEATURE_DIR and available docs, then inspects checklists and halts or prompts if any checklist is incomplete. It loads tasks.md and plan.md (plus optional artifacts like data-model.md, contracts/, research.md, quickstart.md) to build a phase-by-phase execution plan, validates project ignore files, and executes tasks respecting dependencies and parallel markers. Progress, failures, and file changes are reported continuously, and completed tasks are marked off in tasks.md.

When to use it

• You have a complete plan.md and tasks.md and are ready to implement.
• You want automated, traceable execution of feature implementation steps.
• You need checklist enforcement before implementation proceeds.
• You want consistent creation/verification of ignore files based on tech stack.
• You need phase-gated execution with TDD-style test-first task ordering.

Best practices

• Ensure tasks.md and plan.md are complete and up to date before running.
• Run speckit-tasks first if tasks are missing or incomplete.
• Keep checklists in specs/<feature>/checklists/ current to avoid unexpected halts.
• Commit or stash unrelated work to keep task-based changes atomic.
• Review suggested ignore file patterns from the plan and confirm project specifics.

Example use cases

• Implementing a new API feature where plan.md defines tech stack and tasks.md lists TDD steps.
• Onboarding an engineered feature with checklists for UX, security, and tests that must pass before coding.
• Automating repository hygiene by generating .gitignore, .dockerignore, and other ignore files aligned to detected technologies.
• Running phase-by-phase work for a complex feature requiring sequential and parallel tasks with file-based coordination.
• Validating completion by auto-marking tasks as done and producing a final implementation summary.