planner
Interactive planning and execution for complex tasks. Use when breaking down multi-step projects (planning) or executing approved plans through delegation (execution). Planning creates milestones with specifications; execution delegates to specialized agents.
- License 许可
- MIT
- 最近更新
- 2026-03-11
- Stars
- ★ 0
- Forks
- 0
01 前言 Front Matter
- name
- planner
- model
- claude-opus-4-5
- agents
- [ "planner", "architect", "implementer" ]
- license
- MIT
- metadata
- { "version": "1.0.0", "timelessness": "high" }
- description
- Interactive planning and execution for complex tasks. Use when breaking down multi-step projects (planning) or executing approved plans through delegation (execution). Planning creates milestones with specifications; execution delegates to specialized agents.
02 资源文件 7
- 参考文档references/DEVELOPMENT.md3.9 KB
- 参考文档resources/default-conventions.md3.8 KB
- 参考文档resources/diff-format.md7.9 KB
- 参考文档resources/plan-format.md6.1 KB
- 参考文档resources/temporal-contamination.md6.0 KB
- 脚本scripts/executor.ts22.3 KB
- 脚本scripts/planner.ts22.8 KB
03 SKILL.md 正文
Planner Skill
Purpose
Two workflows for complex tasks:
- Planning workflow (planner.ts): Create and review implementation plans
- Execution workflow (executor.ts): Execute approved plans through delegation
Invocation Routing
Invoke planner.ts when user asks to:
- "plan", "design", "architect" a feature
- "review" an existing plan
- Break down a complex task into milestones
Invoke executor.ts when user asks to:
- "execute", "implement", "run" a plan
- "resume" or "continue" execution
- Provides a plan file path for implementation
When to Use
Use the planner skill when the task has:
- Multiple milestones with dependencies
- Architectural decisions requiring documentation
- Migration steps that need coordination
- Complexity that benefits from forced reflection pauses
When to Skip
Skip the planner skill when the task is:
- Single-step with obvious implementation
- A quick fix or minor change
- Already well-specified by the user
Scripts
| Script | Purpose |
|---|---|
scripts/planner.ts |
Planning and review workflow with step-based state management |
scripts/executor.ts |
Execution workflow for approved plans with milestone delegation |
Triggers
| Trigger Phrase | Operation |
|---|---|
plan this feature |
planner.ts (planning phase) |
create implementation plan |
planner.ts (planning phase) |
review the plan and pick up next item |
executor.ts (execution phase) |
execute the plan at plans/X.md |
executor.ts (execution phase) |
resume execution |
executor.ts (continue from last step) |
Anti-Patterns
| Avoid | Why | Instead |
|---|---|---|
| Skipping review phase after planning | Misses quality/temporal issues | Always run review steps 1-2 before execution |
| Starting execution without /clear | Context pollution from planning | User should /clear before execution workflow |
| Manually following workflow steps | Script manages state and transitions | Run the script and follow its output |
| Planning single-step tasks | Overhead exceeds benefit | Implement directly without planner |
| Editing plan during execution | Creates drift between plan and actions | Return to planning phase for changes |
Verification
After planning:
- [ ] Plan file written to specified path
- [ ] Review phase completed (both TW and QR steps)
- [ ] Review verdict is PASS or PASS_WITH_CONCERNS
After execution:
- [ ] All milestones marked complete
- [ ] Post-implementation QR passed
- [ ] Documentation step completed
- [ ] Retrospective generated
Process
Planning Overview
PLANNING PHASE (steps 1-N)
|
v
Write plan to file
|
v
REVIEW PHASE (steps 1-2)
|-- Step 1: @agent-technical-writer (plan-annotation)
|-- Step 2: @agent-quality-reviewer (plan-review)
v
APPROVED --> Execution workflow
Planning Preconditions
Before invoking step 1, you MUST have:
- Plan file path - If user did not specify, ASK before proceeding
- Clear problem statement - What needs to be accomplished
Planning Invocation
bun run ${CLAUDE_SKILL_DIR}/scripts/planner.ts \
--step-number 1 \
--total-steps <estimated_steps> \
--thoughts "<your thinking about the problem>"
Planning Arguments
| Argument | Description |
|---|---|
--phase |
Workflow phase: planning (default) or review |
--step-number |
Current step (starts at 1) |
--total-steps |
Estimated total steps for this phase |
--thoughts |
Your thinking, findings, and progress |
Planning Steps
- Confirm preconditions (plan file path, problem statement)
- Invoke step 1 immediately
- Complete REQUIRED ACTIONS from output
- Invoke next step with your thoughts
- Repeat until
STATUS: phase_complete - Write plan to file using format below
Phase Transition: Planning to Review
When planning phase completes, the script outputs an explicit ACTION REQUIRED
marker:
============================================
>>> ACTION REQUIRED: INVOKE REVIEW PHASE <<<
============================================
You MUST invoke the review phase before proceeding to execution.
The review phase ensures:
- Temporally contaminated comments are fixed (via @agent-technical-writer)
- Code snippets have WHY comments (via @agent-technical-writer)
- Plan is validated for production risks (via @agent-quality-reviewer)
- Documentation needs are identified
Review Phase
After writing the plan file, transition to review phase:
bun run ${CLAUDE_SKILL_DIR}/scripts/planner.ts \
--phase review \
--step-number 1 \
--total-steps 2 \
--thoughts "Plan written to [path/to/plan.md]"
Review Step 1: Technical Writer
Delegate to @agent-technical-writer with mode: plan-annotation
Review Step 2: Quality Reviewer
Delegate to @agent-quality-reviewer with mode: plan-review
After Review
- PASS / PASS_WITH_CONCERNS: Ready for execution workflow
- NEEDS_CHANGES: Return to planning phase to address issues
Execution Workflow (executor.ts)
Execution Overview
Step 1: Execution Planning
|
v
Step 2: Reconciliation (conditional, if prior work signaled)
|
v
Step 3: Milestone Execution (repeat until all complete)
|
v
Step 4: Post-Implementation QR
|
v
QR issues? --YES--> Step 5: Issue Resolution --> delegate fixes --> Step 4
|
NO
v
Step 6: Documentation
|
v
Step 7: Retrospective
Execution Preconditions
Before invoking step 1, you MUST have:
- Approved plan file - Plan that passed review phase
- Clear context window - User should /clear before execution
Execution Invocation
bun run ${CLAUDE_SKILL_DIR}/scripts/executor.ts \
--plan-file PATH \
--step-number 1 \
--total-steps 7 \
--thoughts "<user's request and context>"
Execution Arguments
| Argument | Description |
|---|---|
--plan-file |
Path to the approved plan file |
--step-number |
Current step (1-7) |
--total-steps |
Always 7 for executor |
--thoughts |
Your current thinking and status |
Execution Steps
| Step | Name | Purpose |
|---|---|---|
| 1 | Execution Planning | Analyze plan, detect reconciliation, strategy |
| 2 | Reconciliation | (conditional) Validate existing code vs plan |
| 3 | Milestone Execution | Delegate to agents, run tests (repeat) |
| 4 | Post-Implementation QR | Quality review of implemented code |
| 5 | Issue Resolution | (conditional) Present issues, collect fixes |
| 6 | Documentation | TW pass for CLAUDE.md, README.md |
| 7 | Retrospective | Present execution summary |
Note: Step 3 may be re-invoked multiple times until all milestones complete. Step 4 may loop back through step 5 until QR passes.
Resources
| Resource | Purpose |
|---|---|
resources/plan-format.md |
Plan template (injected at planning completion) |
resources/diff-format.md |
Authoritative specification for code change format |
resources/temporal-contamination.md |
Detecting/fixing temporally contaminated comments |
resources/default-conventions.md |
Default conventions when project docs are silent |
Note: Execution guidance is embedded directly in scripts/executor.ts (not in
separate resource files) since it's only used by that script.
Quick Reference
# === PLANNING WORKFLOW ===
# Start planning
bun run ${CLAUDE_SKILL_DIR}/scripts/planner.ts --step-number 1 --total-steps 4 --thoughts "..."
# Continue planning
bun run ${CLAUDE_SKILL_DIR}/scripts/planner.ts --step-number 2 --total-steps 4 --thoughts "..."
# Start review (after plan written)
bun run ${CLAUDE_SKILL_DIR}/scripts/planner.ts --phase review --step-number 1 --total-steps 2 \
--thoughts "Plan at plans/feature.md"
# Continue review
bun run ${CLAUDE_SKILL_DIR}/scripts/planner.ts --phase review --step-number 2 --total-steps 2 \
--thoughts "TW done, ready for QR"
# === EXECUTION WORKFLOW ===
# Start execution
bun run ${CLAUDE_SKILL_DIR}/scripts/executor.ts --plan-file plans/feature.md --step-number 1 \
--total-steps 7 --thoughts "Execute the feature plan"
# Continue milestone execution
bun run ${CLAUDE_SKILL_DIR}/scripts/executor.ts --plan-file plans/feature.md --step-number 3 \
--total-steps 7 --thoughts "Completed M1, M2. Executing M3..."
# After QR passes
bun run ${CLAUDE_SKILL_DIR}/scripts/executor.ts --plan-file plans/feature.md --step-number 6 \
--total-steps 7 --thoughts "QR passed. Running documentation."
# Generate retrospective
bun run ${CLAUDE_SKILL_DIR}/scripts/executor.ts --plan-file plans/feature.md --step-number 7 \
--total-steps 7 --thoughts "Execution complete. Generating retrospective."
04 同仓库更多 Skills
validation-authority
Treat upstream validators as authoritative. Align local config to them. Use when validation fails unexpectedly, before modifying validator behavior, or when tempted to change upstream tool code.
using-brain-memory
Guidance for using Brain semantic memory effectively. Applies Zettelkasten atomic note principles with observations and relations. Use when deciding whether to query or create notes, structuring note content, or understanding the semantic knowledge graph.
threat-modeling
Structured security analysis using OWASP Four-Question Framework and STRIDE methodology. Generates threat matrices with risk ratings, mitigations, and prioritization. Use for attack surface analysis, security architecture review, or when asking what can go wrong.
taste-lints
Custom lints with agent-readable remediation instructions. Enforces taste invariants (file size, naming conventions, structured logging, complexity) and surfaces errors that agents can act on directly. Use when writing or reviewing code to catch style violations early.
style-enforcement
Validate code against style rules from .editorconfig, StyleCop.json, and Directory.Build.props. Detects line ending violations, naming convention issues, indentation problems, and charset mismatches across C#, Python, TypeScript, and JavaScript. Produces JSON reports for pre-commit hooks and CI pipelines.
steering-matcher
Match file paths against steering file glob patterns to determine applicable steering guidance. Use when orchestrator needs to inject context-aware guidance based on files being modified.