name: reset-story description: Safely reset a blocked or failed Evoloop story back to the build stage — clears deploy attempts and failure notes while preserving verification data argument-hint: "<US-XXX>" disable-model-invocation: true allowed-tools: - Read - Glob - Bash(jq *) - Bash(mv *) - Write

Story Reset Helper

Input

Story ID from $ARGUMENTS. If empty → print usage. If prd/$ARGUMENTS.json missing → list available stories.

Pre-Reset

Read and display: title, stage, deploy attempts/notes, validation data. If complete → warn about full re-run. If build with 0 attempts → "No reset needed", stop.

Confirm

Ask: "Reset <ID> from '<stage>' to 'build'? Clears deploy attempts and notes."

Reset

bash

jq '.stage = "build" | .status.deploy.attempts = 0 | .status.deploy.notes = "" | .status.deploy.passes = false | .status.build.passes = false | .status.build.notes = "" | .status.review.passes = false | .status.review.notes = ""' prd/<ID>.json > prd/<ID>.json.tmp.$$ && mv prd/<ID>.json.tmp.$$ prd/<ID>.json

Critical: .stage is top-level. Preserve .status.validation entirely.

Post-Reset

Display old vs new stage, cleared attempts, preserved validation. Suggest: ./orchestrator.sh run --tool claude --story <ID>