Skip to main content

planning

Planning and orchestration patterns for turning messy context into an incremental, verifiable execution plan. Use when you need a step-by-step strategy before coding or delegating to subagents.

Aller à l'installation

Informations de source

Dépôt
whitebeardit/.cursor
Dernière activité de la source
2 mars 2026 à 02:17
Langue détectée de SKILL.md
anglais
Étoiles
12
Forks
6

Options d'installation

Le prompt qui vérifie d'abord la source est sélectionné par défaut. Vous pouvez passer à une commande directe ou télécharger une copie locale.

Vérifiez les fichiers source

Lisez SKILL.md et les fichiers associés affichés par SkillsMP avant de décider de l'installer.

Affichage de SKILL.md

SKILL.md
Instructions source · Aperçu en lecture seule
name
planning
description
Planning and orchestration patterns for turning messy context into an incremental, verifiable execution plan. Use when you need a step-by-step strategy before coding or delegating to subagents.
# Planning & Orchestration This skill helps you transform an ambiguous or messy situation into a **small-steps execution plan** with clear verification criteria, aligned with the repository's conventions. ## When to Use **✅ DO Use:** - When the user asks for a plan/strategy instead of code - When requirements are unclear or the scope is large - When you need to sequence tasks (tests → refactor → security → docs) - When coordinating multiple subagents (testing, refactor, security, documentation) - When you must preserve constraints (architecture rules, contracts, dependencies) **❌ DON'T Use:** - Trivial tasks with 1–2 obvious steps - When the user explicitly asked for implementation right now (skip planning and implement) - When you already have a validated plan and only need execution ## Core Concepts ### 1) Constraints First Capture non-negotiables before proposing steps: - Tech stack/runtime constraints (Node/TS, frameworks) - Architecture rules (layer boundaries, repositories required, etc.) - Contract-first requirements (OpenAPI, runtime validation) - Observability/security baselines (logging, correlation id, rate limit, helmet) ### 2) Incremental Delivery (Micro-iterations) Prefer **small, verifiable increments** over big-bang rewrites. Each iteration should have: - A single goal - Minimal change set - A verification step (tests, lint, manual smoke check) - A clear rollback story (what to revert if it fails) ### 3) Verification Is Part of the Plan Every planned step must have at least one verification method: - **Automated**: unit/integration tests, contract validation, linting - **Manual**: curl/Postman checks for a narrow endpoint behavior - **Safety**: ensure sensitive fields (e.g., passwords) never leak ## Planning Patterns ### Pattern 1: Situation Snapshot In 5–10 lines, summarize: - What exists today (endpoints, modules, structure) - What is broken/risky (security holes, missing validation, lack of tests) - What must not change (API contract, core behavior, compatibility) ### Pattern 2: Gap-to-Goal Mapping List goals and map each to a concrete deliverable: - **Tests** → test files + scenarios + data builders + how to run - **Architecture** → target folder structure + boundaries + dependency injection approach - **Security** → middleware + validation + secrets/token strategy + error handling ### Pattern 3: Micro-Iteration Plan (Ralph Loop) Produce steps as **small loops**: - Goal - Changes (specific files/modules) - Risks - Verification - Next-step prompt (optional) to delegate to a specialist subagent ### Pattern 4: “Next Prompts” (Delegation) When appropriate, propose copy-pastable prompts to delegate execution: - `/agt-dev-test` for tests - `/agt-dev-refactor` for layering/refactors - `/agt-dev-security` for hardening ## Output Template When delivering a plan, use this structure: 1. **Summary** 2. **Constraints & Assumptions** 3. **Current Risks / Findings** 4. **Plan (small steps)**: - Step name - Goal - Changes (files/modules) - Verification 5. **Suggested Next Prompts** (optional, copy/paste) ## Key Principles 1. **Small steps beat perfect steps** 2. **Name files and verification explicitly** 3. **Respect architecture boundaries and contracts** 4. **Security and observability are not “later”**
Voir sur GitHub