Stats
Actions
Tags
From dl
Plans features as task lists, decomposes work into ordered steps, and saves plan artifacts. Use when planning, decomposing, or organising work into tasks.
How this skill is triggered — by the user, by Claude, or both
Slash command
/dl:planThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
Execute each section in order. Copy the checklist and check off items as you complete them. Do not proceed past a **Gate** until verified.
Execute each section in order. Copy the checklist and check off items as you complete them. Do not proceed past a Gate until verified.
Artifact: Write to .work/plans/. Verify with ls before wrapping up.
Constraint: Do not write code. This skill produces a plan, not an implementation.
Check .work/active/ for marker files. If exactly one exists and no $ARGUMENTS provided, auto-select that feature's slug. If multiple markers exist, list them and ask. Arguments always override.
If $ARGUMENTS (or auto-selected slug) matches a file in .work/plans/ → refine mode. If ambiguous, list and ask. Set but no match → create mode. Empty → list .work/plans/; none: ask what to plan; one: offer refine or new; many: numbered list, ask to pick.
Task Progress:
- [ ] Determine mode
- [ ] If refine: read current plan, check for new brainstorm/research artifacts
- [ ] Gather context: CLAUDE.md, project scan, brainstorm/research artifacts
- [ ] If greenfield: run /resolve-rules scope:bootstrap for stack detection
- [ ] List context and ask clarifying questions (only unresolved items)
- [ ] Create tasks with TaskCreate
- [ ] Write plan to .work/plans/
- [ ] Gate: verify plan with ls; update active marker
- [ ] Present plan for spot-check
Read CLAUDE.md if present. Scan project directory structure. Check .work/brainstorms/ and .work/research/ for matching artifacts — read them if found. If research contains linked repo context, note it. If greenfield (no CLAUDE.md, near-empty directory), run /resolve-rules mode:all scope:bootstrap — if stack rule found, add scaffold as task 1.
List what you found before asking. Ask 3-5 clarifying questions. Read decisions from brainstorm and recommendations from research — only ask about genuinely unresolved items. Wait for answers.
Create tasks with TaskCreate. Decompose by user-visible behavior or capability, not by architectural layer. Do not create tasks like "set up all database models," "build the service layer," or "add tests for everything" — each task should touch all relevant layers for its slice. Order by dependency, then by what unblocks the most follow-up work.
Check for new brainstorm/research artifacts. Show the current plan. Ask "What would you like to change?" Iterate until confirmed. Write the updated file once.
# <Feature Name>
> One sentence: what this feature does and why.
## Tasks
- [ ] **Task title** — what it does; key constraints or dependencies if any.
- **Done when:** one-line acceptance criterion.
- **Verified by:** how to prove this slice works (test, command, or observable behavior).
Verify plan with ls; write if missing. Create or update .work/active/<slug>.md with stage: plan and today's date; derive slug from feature description if creating (lowercase, hyphens, strip non-alphanumeric).
Present the plan. Suggest the user spot-check task ordering, then run /dl:design.
npx claudepluginhub minusblindfold/devloop --plugin dlPlans tasks or features: loads project context including monorepo checks, clarifies requirements, spawns Plan agent, persists validated plans to .groundwork-plans/.
Creates structured PLAN.md for software engineering tasks by interviewing user for intent, scanning project context, and outlining scope, success criteria, sequenced checklist. Use before implementation.
Generates executable Markdown implementation plans for multi-step tasks from context briefs, resolving ambiguities, ordering dependencies, and enabling parallel worker execution.