Agent skill

planning

Research-backed implementation planning skill. This skill should be used when the /plan command is executed, when creating implementation plans, roadmaps, or architectural designs, or when tasks have high uncertainty requiring research before coding. Transforms user requests into execution-ready plans (PRP) using R.P.I.V workflow (Research → Plan → Implement → Validate).

Stars 163
Forks 31

Install this agent skill to your Project

npx add-skill https://github.com/majiayu000/claude-skill-registry/tree/main/skills/other/other/planning-grupous-gpus-3

SKILL.md

Planning Skill

Research-first planning methodology that eliminates unknowns before implementation.

Core Principle: Context Density > Brevity | Research First > Implementation | Planning > Coding

Activation Triggers

This skill is mandatory when:

  1. User executes /plan command
  2. Building plans, roadmaps, or architecture for new features/systems
  3. High uncertainty or risk of hallucination without research
  4. Multi-step execution requiring task decomposition
  5. Third-party integrations (APIs, frameworks, infrastructure)

Skip for: Simple Q&A, pure copywriting, tasks solvable from provided context.


R.P.I.V Workflow

Phase 0: RESEARCH (Always First)

Eliminate unknowns and lock in best-practice approach.

Priority order:

  1. Search codebase for patterns, conventions
  2. Query Context7 for official docs
  3. Tavily web search for best practices
  4. Sequential Thinking for complex decisions

Required outputs:

  • Findings Table: | # | Finding | Confidence (1-5) | Source | Impact |
  • Knowledge Gaps: What remains unknown
  • Assumptions to Validate: Explicit assumptions needing confirmation
  • Edge Cases: Minimum 5 for L4+ complexity

See references/mcp-usage.md for detailed MCP tool guidance.

Phase 1: PLAN (Before Implementation)

Convert research into execution runbook with atomic tasks.

Each task must have:

  • Clear action verb + specific target
  • Validation command
  • Rollback steps
  • Dependencies mapped
  • Mark ⚡ PARALLEL-SAFE when independent

Output: docs/PLAN-{task-slug}.md

See references/plan-template.md for complete template structure.

Phase 2: IMPLEMENT (Only If Requested)

Execute per atomic tasks with validation gates.

Pattern: Implement → Validate → Commit (or Rollback)

Quality gates after each task:

  • bun run check (TypeScript)
  • bun run format (Prettier)
  • bun test (Unit tests)

Phase 3: VALIDATE (Always)

  • Build: zero errors
  • Lint: zero warnings
  • Tests: all passing
  • Consult specialists if security or schema changes involved

Operating Modes

Mode Behavior Output
CONSERVATIVE (default for /plan) Research + plan only, no code docs/PLAN-{slug}.md
PROACTIVE (for implementation) Research → plan → implement Code + plan file

Complexity Levels

Level Indicators Research Depth
L1-L2 Bug fix, single function Repo-only
L3-L5 Feature, multi-file Docs + repo
L6-L8 Architecture, integration Deep research
L9-L10 Migrations, multi-service Comprehensive

See references/complexity-guide.md for detailed classification.


Anti-Patterns

Bad Good
"Implement auth" Research → Search codebase → Query docs → Then implement
Skip research ALWAYS research first, even for "simple" tasks
Guess file paths Search and verify paths before referencing
Speculate about code Read files before making claims

Quick Reference

R.P.I.V: RESEARCH → PLAN → IMPLEMENT → VALIDATE

GOLDEN RULES:
✓ RESEARCH FIRST — never implement blind
✓ Be EXPLICIT — agent follows literally
✓ Explain WHY — enables generalization
✓ ATOMIC TASKS — small, validated, rollback-ready
✓ REFLECT AFTER TOOLS — think before next action

Post-Planning Message

After creating the plan file:

✅ Plan created: docs/PLAN-{slug}.md

Next steps:
- Review the plan
- Run `/implement` to start implementation
- Or modify plan manually

Resources

  • references/plan-template.md — Complete PRP template structure
  • references/complexity-guide.md — Task complexity classification
  • references/mcp-usage.md — When and how to use MCP tools

Expand your agent's capabilities with these related and highly-rated skills.

Didn't find tool you were looking for?

Be as detailed as possible for better results