How this skill is triggered — by the user, by Claude, or both
Slash command
/spwnr:spwnr-planThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Use this skill for `/spwnr-plan`. It creates or revises the latest active plan artifact, then asks whether to execute, improve, or stop. Use `spwnr-principle` for shared contracts: plan artifact convention, execution units, task metadata, approval gates, worker readiness, and the execution review loop.
Use this skill for /spwnr-plan. It creates or revises the latest active plan artifact, then asks whether to execute, improve, or stop. Use spwnr-principle for shared contracts: plan artifact convention, execution units, task metadata, approval gates, worker readiness, and the execution review loop.
spwnr-principle plus this skill.TodoWrite for blockers, readiness fields, and the active plan path.Write or targeted Edit; do not leave the durable plan only in chat.AskUserQuestion only for decisions that change decomposition, sequencing, acceptance criteria, or execution topology.Agent passes are allowed only after a draft exists and only for sequential research, draft, and review.Agent only for research, draft, and review.Metadata, Locked Readiness Fields, Detailed Plan, Execution Units, Expert Planning Round, and Plan Review Loop.A ready plan must be execution-grade and fresh-worker-safe. It must lock goal, success criteria, boundaries, constraints, assumptions, risks, active artifact path, execution strategy, executable units, capability requirements, failure rules, and approval condition.
Reject thin plans that restate the request, use vague verbs, hide sequencing behind as needed, or combine materially different actions only to stay short. At least one rejected approach must have a concrete operational reason.
spwnr-principle.Use these sections in order:
MetadataUser RequestLocked Readiness FieldsApproach AnalysisDetailed PlanExpert Planning RoundDecisions NeededPlan Review LoopPending Handoff NotesDetailed Plan must include:
Mission Layer: goal, success criteria, in scope, out of scope, non-goals, constraintsExecution Architecture: dependencies, ownership, approval boundaries, parallel-safety assumptions, serialization pointsEnvironment And PreconditionsExecution Strategy Recommendation: selected pipeline or team, rationale, and pattern shapeAgent Capability Requirements: generic capability needs by unit and, for pipeline, by stageExecution Units: fields from spwnr-principle, including risk level, file ownership hints, and worker plan approval. Prefer - **unit_id**: unit-1.Failure And Escalation RulesFor pipeline, persist pattern name, ordered stages, handoffs, and acceptance checks. For team, persist whether to launch multiple pipelines in parallel or use one shared queue, and document any shared-file exception.
Decisions Needed should contain only unresolved decisions or locked defaults. Pending Handoff Notes should contain only execution-relevant reminders.
Plan metadata must include Revision Status, Superseded By, material re-plan handling, and the latest active revision.
Run research -> draft -> review only when it materially improves the plan.
spwnr resolve-workers briefs.Worker Readiness Required and preserve the active revision.Expected contributions:
research: missing constraints, dependencies, assumptions, risksdraft: thin-section rewrites and executable unit improvementsreview: ambiguity, coupling, missing approvals, sequencing gaps, weak acceptance checksBefore finalizing, improve the 3 weakest sections: split bundled units, replace vague verbs with artifact-producing actions, make acceptance checks local/objective, and verify ownership plus escalation still match the chosen execution strategy.
After each write or revision, record Plan Review Loop and ask for Execute current plan, Continue improving plan, or End this round. Do not recreate the old needs-confirmation or approved-plan-ready state machine in the plan file.
Default sections: Plan Artifact, Detailed Plan, Next Step.
Expose planning internals only when useful or requested. Do not present a plan that is merely formatted correctly; present one that can be handed to a fresh worker with minimal interpretation.
Guides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.
npx claudepluginhub cheong43/spwnr --plugin spwnr