From claude-meta-tools
Rigorous planning workflow with research validation, task breakdown, dependency mapping, and post-implementation verification. Use when: (1) planning complex features touching 5+ files, (2) architectural changes or major refactors, (3) unfamiliar tech requiring doc research, (4) user says "thorough plan", "rigorous plan", or "structured plan". Invoke with /structured-plan before or during plan mode.
How this skill is triggered — by the user, by Claude, or both
Slash command
/claude-meta-tools:structured-planThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
A checklist-driven planning workflow that ensures plans are research-validated, broken into atomic tasks, complete, and self-verifying.
A checklist-driven planning workflow that ensures plans are research-validated, broken into atomic tasks, complete, and self-verifying.
Phase A: Clarification (if needed)
↓
Phase B: Iterative Refinement (7 steps)
↓
Exit Plan Mode → Implementation
If the user's prompt is vague or ambiguous, use AskUserQuestion to clarify before proceeding. Ask about:
Once intent is clear, proceed to Phase B.
After drafting an initial plan, apply this checklist. Edit the plan file after each step (not one-shot).
Validate your approach against official documentation.
mcp-add context7 (lean mode) or skip to WebSearch## Best Practices References
- [Pattern name](url) - key insight
- [Library docs](url) - relevant section
→ Edit plan file
Break the plan into atomic, committable tasks.
See references/task-template.md for format.
→ Edit plan file
Define execution order with a dependency diagram.
→ for sequential dependenciesTask 1 → Task 2 → Task 3
Task 4, Task 5 (parallel, after Task 3)
→ Edit plan file
Identify code that would be duplicated across tasks and define shared locations.
Add a Shared Infrastructure section listing where shared code lives.
→ Edit plan file
Ensure the plan is unambiguous enough for another agent to implement.
... or "rest unchanged" placeholdersTest: Could someone implement this without asking questions?
→ Edit plan file (fix any gaps)
Add a final task that spawns 3 parallel verification subagents after implementation.
See references/verification-agents.md for prompts.
### Task N: Post-Implementation Verification
**Files**: None (verification only)
**Verify**: Spawn 3 parallel subagents:
1. Compliance Agent (Explore) - verify implementation matches plan
2. Best Practices Agent (general-purpose) - validate against docs
3. Code Simplifier Agent (code-simplifier) - check for over-engineering
**Commit**: None (verification task)
→ Edit plan file
| Step | Action | Plan Section Added |
|---|---|---|
| 1 | Research | Best Practices References |
| 2 | Task breakdown | Implementation Tasks |
| 3 | Dependencies | Task Dependencies diagram |
| 4 | Shared code | Shared Infrastructure |
| 5 | Completeness | (fixes throughout) |
| 6 | Verification | Final verification task |
| 7 | Exit | (tool call) |
npx claudepluginhub bengous/claude-code-plugins --plugin claude-meta-toolsGenerates detailed, step-by-step implementation plans from specs or requirements, with bite-sized tasks, file paths, and test-first guidance.
Generates executable Markdown implementation plans for multi-step tasks from context briefs, resolving ambiguities, ordering dependencies, and enabling parallel worker execution.
Creates detailed implementation plans through interactive, iterative collaboration. Use when planning features, refactoring, or tasks with research and multi-agent analysis.