Implement

Finally, in Implement, the AI carries out the tasks:

• Writing tests

• Implementing production code

• Iterating until requirements are satisfied

Human review and guidance remain part of the loop, but the AI performs the bulk of the mechanical execution.

✏️ Steps

• Run:

  • /speckit.implement

    • Open

      

  • What happens:

    • The agent validates prerequisites (constitution/spec/plan/tasks), and parses tasks.md, executes tasks in order, follows the TDD approach defined in the tasks, and reports progress/errors.

  • Important caveat:

    • Implementation often involves running local CLI commands (npm/dotnet/etc.), so your dev environment needs the required tools installed.

• Practical workflow:

  • If tests fail or you hit runtime errors (including browser console errors), paste the error output back into the agent and let it iterate until green.

Next Steps

At this point you should have a finished feature! Now it’s time to test the feature and work with the Agent to fix any mistakes

Feature Development Workshop | 8. Manually test the new featurePreview