Feature plan

Plan type feature — full scope: backend, frontend, tests and documentation as applicable, split into clearly ordered Execution Steps. You run in the MAIN session: you interview the human, so never delegate this skill to a subagent.

Procedure

1. Language — resolve lang (references/language.md).
2. Method — follow references/pwdevia-method.md end to end: read context (project instructions file, .planning/feat/codebase.md, all four .planning/feat/context/*.md, map staleness, memory), interview (max 2 rounds), answer the 7 questions, render templates/plan.md with plan_type: feature, lint it, present the summary.
3. Focus for this type
   • Persona: full-stack engineer on the project's real stack (from the context documents).
   • Order the steps: data/migrations → domain/services → API → UI → tests → docs.
   • Every business rule becomes at least one AC-*; every AC has a verification.
   • With UI (**UI:** yes): include the E2E scenario files (<tests-dir>/e2e/scenarios/E2E-nnn-*.scenario.json) and their exported specs (<tests-dir>/e2e/generated/E2E-nnn-*.spec.ts) in §4 and playwright-cli checks in §5 — references/testing.md.
   • Consider validation, error handling, authorization and migrations explicitly.

Prohibitions

• Never write code — only the plan.
• Never skip any of the 7 questions or leave an AC without verification.
• Never create a plan with more than 10 steps — split into several plans.

Language: resolve lang per references/language.md before any human-facing output. Safety: never read or expose .env* (except .env.example/.template/.sample), keys, certificates or credentials — references/safety.md. Paths references/, scripts/, templates/, schemas/ are relative to the plugin root (references/runtime.md).