goal-plan-clean

goal-plan-clean

Compose a /goal that drives the plan review-fix loop:

1. Run /lets:review --plan (or /lets:check --plan)
2. Read findings from .lets/reviews/<latest>
3. Edit the plan file to address each BLOCKER and SUGGESTION
4. Re-run review/check
5. Repeat ≤3 iterations until verdict APPROVED

The skill never invokes /goal itself — Claude Code's /goal is a UI command (verified empirically: Skill(skill: "goal") returns "goal is a UI command, not a skill. Ask the user to run /goal themselves"). It composes well-structured text and presents for the user to paste.

IMPORTANT: If the spec below invokes any deferred tool (e.g. AskUserQuestion), you MUST load and call it as specified. Never skip the call, never substitute a default answer of your own — the tool invocation is part of the contract. This is critical.

Step 1: Resolve plan file

LETS_PROJECT_ROOT=$(git rev-parse --show-toplevel)
BRANCH=$(git branch --show-current)
SLUG=${BRANCH#feature/}
PLAN_PATH="$LETS_PROJECT_ROOT/.lets/plans/${SLUG}.md"
test -f "$PLAN_PATH" && echo "found: $PLAN_PATH" || echo "NOT_FOUND"

If NOT_FOUND:

ls -1 "$LETS_PROJECT_ROOT/.lets/plans/"*.md 2>/dev/null

Then AskUserQuestion to pick from existing plans, or exit if none.

Validation (prevent path traversal): SLUG must match ^[a-zA-Z0-9._-]+$. If not, fail with "Invalid branch name for plan derivation."

Step 2: Resolve review tool

Parse args parameter:

• args = "review" or empty → use /lets:review --plan (deep, multi-agent, slower)
• args = "check" → use /lets:check --plan (inline 6-lens, faster, no agents)

If args is empty AND user context unclear, AskUserQuestion:

AskUserQuestion(
  questions=[{
    question: "Which tool for the iteration loop?",
    header: "Tool",
    options: [
      { label: "review (Recommended)", description: "/lets:review --plan — deep multi-agent review, slower, catches more" },
      { label: "check", description: "/lets:check --plan — inline 6-lens, faster, lighter" }
    ],
    multiSelect: false
  }]
)