Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Create detailed phase plan (PLAN.md) with verification loop
.claude/skills/davepoon-gsd-plan-phase/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-19 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -23% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 48% | 0% |
| case-12 | ✗→✓ | ▲ Improved | 32% | 0% |
| case-13 | ✗→✓ | ▲ Improved | 8% | 0% |
<objective> Create executable phase prompts (PLAN.md files) for a roadmap phase with integrated research and verification.
Default flow: Research (if needed) → Plan → Verify → Done
Orchestrator role: Parse arguments, validate phase, research domain (unless skipped), spawn gsd-planner, verify with gsd-plan-checker, iterate until pass or max iterations, present results. </objective>
<execution_context> @${CLAUDE_PLUGIN_ROOT}/workflows/plan-phase.md @${CLAUDE_PLUGIN_ROOT}/references/ui-brand.md </execution_context>
<runtime_note> Copilot (VS Code): Use vscode_askquestions wherever this workflow calls AskUserQuestion. They are equivalent — vscode_askquestions is the VS Code Copilot implementation of the same interactive question API. Do not skip questioning steps because AskUserQuestion appears unavailable; use vscode_askquestions instead. </runtime_note>
<context> Phase number: $ARGUMENTS (optional — auto-detects next unplanned phase if omitted)
Flags:
--research — Force re-research even if RESEARCH.md exists--skip-research — Skip research, go straight to planning--gaps — Gap closure mode (reads VERIFICATION.md, skips research)--skip-verify — Skip verification loop--prd <file> — Use a PRD/acceptance criteria file instead of discuss-phase. Parses requirements into CONTEXT.md automatically. Skips discuss-phase entirely.--reviews — Replan incorporating cross-AI review feedback from REVIEWS.md (produced by /gsd:review)--text — Use plain-text numbered lists instead of TUI menus (required for /rc remote sessions)Normalize phase input in step 2 before any directory lookups. </context>
<process> Execute the plan-phase workflow from @${CLAUDE_PLUGIN_ROOT}/workflows/plan-phase.md end-to-end. Preserve all workflow gates (validation, research, planning, verification loop, routing). </process>
<output_format> When planning concludes (PLAN.md files written, plan-checker passed, STATE.md updated), emit a Next Up continuation block following the pattern in references/continuation-format.md:
## ✓ Phase N Planned — M plans, K tasks with brief plan list)## ▶ Next Up heading with /gsd:execute-phase N/clear then: /gsd:list-phase-assumptions N, etc.Plan-to-execute is a clean boundary — the discuss/research/plan conversation rarely informs execution. Suggesting /clear here keeps execution starting with a tight, plan-focused context. </output_format>
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-04 | fail→fail | 2,913 | 4,403 | +51% | 1 | 1 | 0% | 315 | 984 | +212% | 0 | 0 | — |
case-19 | fail→pass | 11,118 | 2,218 | -80% | 1 | 1 | 0% | 1,858 | 1,084 | -42% | 0 | 0 | — |
case-01 | fail→fail | 2,502 | 4,259 | +70% | 1 | 1 | 0% | 414 | 932 | +125% | 0 | 0 | — |
case-02 | fail→fail | 5,297 | 5,778 | +9% | 1 | 1 | 0% | 348 | 1,005 | +189% | 0 | 0 | — |
case-03 | fail→fail | 3,945 | 4,867 | +23% | 1 | 1 | 0% | 169 | 965 | +471% | 0 | 0 | — |
case-05 | fail→fail | 6,010 | 4,242 | -29% | 1 | 1 | 0% | 1,017 | 902 | -11% | 0 | 0 | — |
case-06 | fail→fail | 10,755 | 4,952 | -54% | 1 | 1 | 0% | 1,664 | 928 | -44% | 0 | 0 | — |
case-07 | fail→fail | 10,721 | 4,828 | -55% | 1 | 1 | 0% | 1,937 | 914 | -53% | 0 | 0 | — |
case-08 | fail→fail | 9,704 | 5,475 | -44% | 1 | 1 | 0% | 1,626 | 1,021 | -37% | 0 | 0 | — |
case-09 | fail→fail | 11,666 | 4,322 | -63% | 1 | 1 | 0% | 1,633 | 943 | -42% | 0 | 0 | — |
case-10 | fail→pass | 11,036 | 4,357 | -61% | 1 | 1 | 0% | 1,884 | 1,449 | -23% | 0 | 0 | — |
case-11 | fail→pass | 4,673 | 1,904 | -59% | 1 | 1 | 0% | 678 | 1,006 | +48% | 0 | 0 | — |
case-12 | fail→pass | 6,009 | 3,407 | -43% | 1 | 1 | 0% | 1,075 | 1,417 | +32% | 0 | 0 | — |
case-13 | fail→pass | 6,507 | 2,751 | -58% | 1 | 1 | 0% | 1,015 | 1,094 | +8% | 0 | 0 | — |
case-14 | fail→pass | 10,075 | 2,277 | -77% | 1 | 1 | 0% | 1,701 | 1,102 | -35% | 0 | 0 | — |
case-15 | pass→pass | 5,800 | 3,678 | -37% | 1 | 1 | 0% | 994 | 1,361 | +37% | 0 | 0 | — |
case-16 | fail→pass | 9,563 | 6,372 | -33% | 1 | 1 | 0% | 1,639 | 1,897 | +16% | 0 | 0 | — |
case-17 | fail→pass | 11,049 | 2,940 | -73% | 1 | 1 | 0% | 1,807 | 1,119 | -38% | 0 | 0 | — |
case-18 | fail→pass | 16,737 | 1,648 | -90% | 1 | 1 | 0% | 1,126 | 956 | -15% | 0 | 0 | — |
case-20 | fail→fail | 3,067 | 18,318 | +497% | 1 | 1 | 0% | 515 | 1,165 | +126% | 0 | 0 | — |
case-21 | pass→pass | 5,723 | 11,440 | +100% | 1 | 1 | 0% | 1,024 | 2,286 | +123% | 0 | 0 | — |
case-22 | pass→pass | 8,227 | 9,155 | +11% | 1 | 1 | 0% | 941 | 1,774 | +89% | 0 | 0 | — |
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, and 11 counted toward the lift figure. The other 11 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +41 percentage points is the difference between those two pass rates over the 11 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.