loop-engineeringlisted

# Loop engineering ## Scope and triggers Also fires on "let's build a loop", "loop engineering", "make this a bounded loop", or a prompt/parameter tuning loop. Walks the gate (should this even be a loop) then the anatomy (artifact / scanner / reviser / stop condition), routing to one of three shapes: bounded-revise engine, campaign/worklist, or bounded search-select. ## Overview The kit runs 3 first-party bounded loops: Goal, Debug, Execute. It also runs a bounded-revise side-flow, `test-plan-review-team`. This skill extracts the reusable shape from that fourth loop. Use it to design a new loop against a known pattern, instead of starting from scratch. ## How to use this skill You do not need to design the stop condition or the output shape before you start. Describe the artifact and the itch. This skill runs the rest. **Step 1 needs**: the artifact, and why now. Example: "the README's feature list keeps drifting from the code." This skill checks it against the four gate criteria below and reports pass or fail. **Step 2 needs**: what good and bad look like, in plain words. You do not need to sort checks into Tier 1 or Tier 2 yourself. Describe what wrong looks like, for each check you want. This skill sorts them and shows you the split, so you can correct it. **Step 3 needs**: what the artifact should look like when done. Say whether it is a written-back section, a standalone report, or an applied edit. Say what verdict shape you want. If you do not say, this skill