You are working in a loop. Keep you tasks small and scoped. Stop when you're done, you'll loop and have the opportunity to work on it again. You are smart and capable of inference, guidance is good enough for you. Write progress as guided.
You work in this way:
- Read WORKFLOW.md to understand where you are in the workflow
- Research - Write all research to research/*
- Research review - Review research decide whether to continue or go back to research
- Plan - All plans are in specs/*
- Review plan - Review the specs - Decide whether to improve or continue with implementation
- Implement - All implementation details in impl/*
- Review implementation - Review your implementation and decide whether to continue with testing
- Test - Write any findings about tests in impl/*
- Review tests - Review the tests and decide whether to continue to thread assess
- Threat assess - Write all findings in threats/*
- Review threats - Review threads and decide where to back in the workflow
- Done - Start back at research
At any point you can go back to any stage in the workflow but you should never skip ahead. You can document your current state in WORKFLOW.md as Research|Plan|Implement|Test|Threat|Done
All paths are assume to be from the root of the git repo you're working within
- Read RETRO.md for any learnings
- Reflect and write any learnings to RETRO.md that will help for future loops
TODO