From superpowers-plus
Enforces quality gates when writing plans, roadmaps, or phased work: prevents fabricated timelines, ensures dependency ordering, and requires exit criteria.
How this skill is triggered — by the user, by Claude, or both
Slash command
/superpowers-plus:plan-quality-gatesThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
> **Last Updated:** 2026-03-20
Last Updated: 2026-03-20 Fires alongside:
superpowers:plan-and-execute— this skill is ADDITIVE, not a replacement. Also relevant during brainstorming when plan/roadmap topics arise (load manually if needed).Wrong skill? Executing a plan →
plan-and-execute. Feature design →brainstormingordebate. Writing documentation →readme-authoring. See also: detecting-ai-slop reference § Fabricated Calendar Timelines
Enforce quality constraints on plans at creation time. The upstream plan-and-execute skill handles plan structure. This skill prevents specific failure modes (fabricated timelines, missing exit criteria) that other skills do not guard against.
Announce at start: "Using plan-quality-gates to enforce timeline and exit-criteria discipline."
NEVER assign calendar periods to plan phases without actual capacity data.
You have zero information about:
Therefore you CANNOT write any of the following:
| Forbidden Pattern | Example |
|---|---|
| Phase + calendar period | "Phase 2: Foundation (Weeks 1-2)" |
| Sprint numbering | "Sprint 1: schema extraction" |
| Quarter/month targets | "Target: Q3 2026" |
| Week-numbered milestones | "By Week 3, we should have..." |
| Aggregate duration | "Timeline: 4-6 weeks" |
| Day estimates for phases | "Phase 1 (Days 1-3)" |
If the user provides capacity data (team size, sprint length, velocity), you MAY use it. Otherwise, any calendar-based phasing is fabrication.
Dependency ordering + exit criteria. Each phase states:
Phase 1: Discovery (Week 1)
Phase 2: Foundation (Weeks 1-2)
Phase 3: Validation (Weeks 3-4)
Phase 4: Optimization (Week 5)
Phase 1: Build schema knowledge base
Depends on: nothing
Exit: ≥9 table docs with business semantics committed
Phase 2: Wire named connections
Depends on: Phase 1 (need schema understanding to choose connection names)
Exit: config-service reads from named connections, tests pass
Phase 3: Validate data flow
Depends on: Phase 2
Exit: integration test confirms read path works end-to-end
Phase 4: Cut over production queries
Depends on: Phase 3
Exit: old connection strings removed, monitoring confirms no regressions
If you don't know how long something takes, say "unknown" — don't invent numbers.
| Forbidden | Acceptable |
|---|---|
| "This should take 2-3 days" | "Duration: unknown without team input" |
| "Estimate: 1 sprint" | "Estimate: requires sizing by the team" |
| "Quick win — 1 hour" | "Likely small scope — team to confirm" |
Exception: If the user explicitly asks for rough estimates and accepts they're guesses, you MAY provide them with a clear disclaimer: "⚠️ These are rough guesses, not commitments."
Every phase in a plan MUST state its dependencies explicitly:
Depends on: nothing (can start immediately)Depends on: Phase N (sequential dependency)Depends on: Phase N + external approval (blocked dependency)If two phases have no dependency relationship, note they can run in parallel.
Every phase or task MUST have a concrete "done means" statement:
Exit criteria must be verifiable — someone can check whether they're true without asking the author what they meant.
Before finalizing any plan, verify:
Depends on: statementExit: criterionIf any check fails, fix the plan before presenting it.
detecting-ai-slop integration)These patterns in plan output indicate this skill was not followed:
| Pattern | Score | Severity |
|---|---|---|
| Phase + calendar period | +8 | Fabrication |
| Duration estimate without basis | +5 | Fabrication |
| Missing exit criteria | +3 | Incomplete |
| Missing dependency statement | +3 | Incomplete |
plan-and-execute: Plan writing standardsplan-and-execute skill# Validate plan quality before execution
echo "Checklist:"
echo " [ ] Every step has success criteria"
echo " [ ] Rollback plan documented"
echo " [ ] Dependencies identified"
echo " [ ] No fabricated timelines (capacity data required for any time estimate)"
| Failure | Fix |
|---|---|
| Calendar-disguised phase labels: "Phase 1: Quick Win" implies short duration | Audit for duration-implying words: "quick," "short," "rapid," "brief" |
| Circular exit criteria: "Phase 2 exit: Phase 2 is complete" | Exit criteria must be independently verifiable — "what can someone check?" |
| Parallel phases listed sequentially, implying false dependency chain | Explicitly mark phases that can run in parallel |
| Plan presented without checking all rules | Run the Self-Check (all 5 boxes) before presenting to user |
npx claudepluginhub bordenet/superpowers-plus --plugin superpowers-plusGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.