Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Structured plan file generation for large tasks -- phased planning, completion tracking, and decision documentation
.claude/skills/plan-documentation/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-17 | ✗→✓ | ▲ Improved | — | — |
| case-12 | ✗→✓ | ▲ Improved | — | — |
| case-18 | ✗→✓ | ▲ Improved | — | — |
| case-16 | ✗→✓ | ▲ Improved | — | — |
| case-06 | ✗→✓ | ▲ Improved | — | — |
Generate structured plan files when starting large features, refactoring efforts, or multi-phase work. Plans create a paper trail that survives context resets and helps collaborators understand what was planned, what changed, and why.
Create a plan document when:
Do NOT create plan files for:
Create the plan at plans/<task-slug>-plan.md in the project root.
markdown# Plan: <Task Title> Created: <ISO date> Status: IN_PROGRESS | COMPLETE | ABANDONED ## Summary <2-3 sentences explaining what this work accomplishes and why it matters.> ## Phases ### Phase 1: <Phase Title> - [ ] Step 1 description - [ ] Step 2 description - [ ] Step 3 description **Acceptance Criteria:** - <What must be true when this phase is done> - <Measurable or verifiable condition> ### Phase 2: <Phase Title> - [ ] Step 1 description - [ ] Step 2 description **Acceptance Criteria:** - <Condition> ### Phase N: <Phase Title> ... ## Open Questions 1. <Question about approach or requirement> -- Suggested: <Option A> vs <Option B> 2. <Question about scope> -- Suggested: <Option A> vs <Option B> 3. <Question about dependency or risk> -- Suggested: <Option A> Keep open questions between 1-5. Each should have suggested options so they can be resolved quickly. Remove questions as they get answered -- move the decision to the relevant phase. ## Dependencies - <External system, API, library, or team dependency> - <Prerequisite work that must finish first> ## Risks - <What could go wrong and how likely it is> - <Mitigation strategy if applicable>
After finishing each phase, create plans/<task-slug>-phase-N-complete.md:
markdown# Phase N Complete: <Phase Title> Completed: <ISO date> Plan: <task-slug>-plan.md ## Phase N Summary <1-2 sentences on what was accomplished.> ## What Changed | File | Change | |------|--------| | `path/to/file.ts` | Added validation logic for user input | | `path/to/test.ts` | 6 new unit tests for validation edge cases | ## What Was Tested - <Test suite or manual verification performed> - <Edge cases covered> - <What was NOT tested and why> ## Decisions Made - <Decision>: <Why this option was chosen over alternatives> ## Next Phase Preview Phase N+1 will focus on <brief description>. Prerequisites met: <yes/no>.
When all phases are done, create plans/<task-slug>-complete.md:
markdown# Complete: <Task Title> Completed: <ISO date> Plan: <task-slug>-plan.md Duration: <how long the work took> ## Task Summary <2-3 sentences on the full scope of work completed.> ## All Phases | Phase | Title | Status | |-------|-------|--------| | 1 | <Title> | COMPLETE | | 2 | <Title> | COMPLETE | | N | <Title> | COMPLETE | ## Total Changes - Files modified: <count> - Files created: <count> - Tests added: <count> - Lines changed: <approximate> ## Lessons Learned - <Insight that would help someone doing similar work> - <Unexpected challenge and how it was resolved> - <Pattern discovered that should be reused>
All plan files go in the plans/ directory at the project root.
| File | Purpose | |------|---------| | <task-slug>-plan.md | Initial plan with phases and acceptance criteria | | <task-slug>-phase-N-complete.md | Completion record for phase N | | <task-slug>-complete.md | Final completion summary |
The <task-slug> should be a short, lowercase, hyphenated description of the task (e.g., auth-refactor, search-api-v2, dashboard-redesign).
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-17 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +55 percentage points is the difference between those two pass rates over the 22 comparable cases.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
Other measured skills in the registry, with their headline benchmark lift.