De-Sloppify Pass
FeaturedOn demandAbout De-Sloppify Pass
This loop tidies up the latest changes made by a coding agent so they stay minimal and follow project conventions. Each round reviews the changes, fixes anything superfluous or inconsistent, then runs lint and tests to confirm. It is started manually and runs at most 4 rounds.
Key features
- 01Manual trigger, so the user decides when it starts
- 02Checks each round with npm run lint && npm test
- 03Capped at 4 rounds to avoid endless edits
- 04Supports Claude Code and Cursor
Use cases
- 01Final cleanup after a coding agent finishes a feature, before merging
- 02Large changes where unrelated edits may have slipped in
- 03Keeping new code consistent with existing project conventions
- Goal
- Recent changes are clean, minimal and aligned with project conventions, and lint and tests pass.
- Exit when
- Stop when the review finds nothing left to clean up and npm run lint && npm test passes, or after 4 rounds.
- Between iterations run
- npm run lint && npm test
Launch prompt
Paste into Claude Code / Cursor. The agent repeats the steps below until the exit condition is met.
Start the "De-Sloppify Pass" loop. Goal: Recent changes are clean, minimal and aligned with project conventions, and lint and tests pass. Max iterations: 4. Step 1: Review recent changes — Read through the latest changes and look for superfluous code, unrelated edits and departures from project conventions. Step 2: Clean up findings — Make the smallest edits that resolve what the review found, without refactoring unrelated code. Step 3: Run the checks — Run npm run lint && npm test. Any failure becomes the first thing to address in the next round. `npm run lint && npm test` Step 4: Decide whether to continue — Stop if the review is clean and the checks pass; otherwise start another round within the round limit. Between iterations run: `npm run lint && npm test` Exit when: Stop when the review finds nothing left to clean up and npm run lint && npm test passes, or after 4 rounds.
Steps
Each round runs in order; if the exit condition isn’t met, another round starts, up to 4.
Read through the latest changes and look for superfluous code, unrelated edits and departures from project conventions.
Make the smallest edits that resolve what the review found, without refactoring unrelated code.
Run npm run lint && npm test. Any failure becomes the first thing to address in the next round.
npm run lint && npm testStop if the review is clean and the checks pass; otherwise start another round within the round limit.
Supported agents
Coding agents that can run this loop directly.