| name | understand-propose-implement-verify |
| description | Use when the agent must follow a strict problem-solving loop: understand, propose, implement, verify. |
| license | MIT |
| metadata | {"author":"Petr Král (pekral.cz)"} |
Constraint:
- Read project.mdc file
- First, load all the rules for the cursor editor (.cursor/rules/.*mdc).
- I want the texts to be in the language in which the assignment was written.
- Always follow this order: understand -> propose -> implement -> verify.
- Reuse existing project skills whenever they already solve a phase better than ad-hoc work.
Steps:
- First: understand the problem
- Analyze assignment details, comments, context files, and any linked issue-tracker resources.
- Classify the request (bug, feature, refactor, review, docs, infra).
- Build a short task checklist with assumptions and constraints.
- Then: propose solution
- Propose the smallest safe solution that satisfies the request.
- Explain expected impact, trade-offs, risks, and why this approach is preferred.
- Select and invoke relevant existing skills for the task (e.g.
resolve-github-issue, create-test, code-review, security-review, process-code-review).
- Then: implement
- Execute the proposed solution end-to-end.
- Keep changes focused, deterministic, and aligned with existing project conventions.
- Add or update tests for all changed behavior.
- Finally: verify correctness
- Run required fixers/checkers/tests for changed scope.
- Confirm output quality, regressions, and requirement coverage.
- Report final status with what changed, what was tested, and any remaining risks.
After completing the tasks
- Ensure every response and change can be traced back to this four-step loop.
- Ensure existing skills were reused where applicable instead of duplicating logic.
Output Humanization
- Use blader/humanizer for all skill outputs to keep the text natural and human-friendly.
