The capabilities section describes the key capabilities for executing plans effectively.
Note: Detailed execution examples are available in separate files to reduce context size. Load only the specific example you need.
Available Examples:
- Single-Component Refactor: Simple 8-step refactor with progress updates
- Load:
examples/single-component-refactor.md
- Multi-File Feature Implementation: Complex 15-step implementation across multiple files
- Load:
examples/multi-file-implementation.md
- Handling Failed Steps: Error recovery and debugging during execution
- Load:
examples/handling-failed-steps.md
When to load: Load a specific example when you need concrete guidance for a similar scenario.
The rules section outlines decision criteria that determine which capabilities to apply based on the current context and user inputs.
At Plan Start: Apply the plan-tracking capability to initialize the tracking system with all plan steps before beginning execution.
During Execution: Apply the step-execution capability continuously:
- Before each step: think aloud and explain your approach
- During each step: execute thoroughly and handle dependencies
- After each step: validate results and update status to completed immediately
For Long or Multi-Phase Plans: Apply the progress-reporting capability:
- After completing major phases (e.g., all tests written, all implementations done)
- After every 8-10 steps for plans with 15+ steps
- When switching contexts (moving to different files/components)
- Always provide a final completion summary at the end
When Steps Fail or Encounter Errors: Apply the error-handling capability immediately:
- Document the error clearly
- Analyze root cause
- Attempt to fix and retry
- If unresolvable, consult the user before proceeding
- Never skip failed validation steps
At Validation Points: Apply the validation-checkpoints capability:
- After code changes: run relevant tests
- After significant changes: run linting, formatting, type-checking
- At major milestones: verify builds succeed
- Don't wait until the end—validate incrementally throughout execution
When Facing Ambiguity or Blockers: Apply the user-interaction capability:
- If a step is unclear: pause and ask for clarification
- If external dependencies are missing: document the blocker and consult the user
- If deviations from the plan are necessary: explain why and how you're adapting
- Otherwise, execute autonomously without asking for permission at each step
Plan Immutability: Throughout execution, apply the plan-tracking immutability principle—never modify the plan structure, objectives, or steps except to update statuses or add brief clarifying notes. Never skip, reorder, or remove steps without explicit user approval.
Complete All Steps: Execute ALL steps in the plan thoroughly, regardless of the number of steps or files affected. Do not stop early or leave steps partially completed.
Converted and distributed by TomeVault — claim your Tome and manage your conversions.
1---2name: plan-executor-23description: Execute an outlined plan step by step, tracking progress and updating the status of each step as you go. Use this skill whenever you need to execute an outlined plan. Use when this capability is needed.4---56<when-to-use-this-skill>7- You need to execute an outlined plan (implementation plan, refactoring plan, or bug-fix plan)8- A structured, multi-step task requires systematic execution and progress tracking9</when-to-use-this-skill>1011<capabilities>1213The capabilities section describes the key capabilities for executing plans effectively.1415<plan-tracking>16- **Initialize Tracking**: At the start of execution, record all plan steps in the tracking system (todo tool or PLAN.md file)17- **Status Updates**: Update step status immediately after completion (not-started → in-progress → completed)18- **Immutable Plan**: Never modify the plan structure, objectives, or steps except to update status or add clarifying notes19- **Preserve Context**: Maintain the original plan's intent, sequence, and dependencies throughout execution20</plan-tracking>2122<step-execution>23- **Sequential Execution**: Execute steps in order unless dependencies allow parallelization24- **Think Aloud**: Before implementing each step, briefly explain your approach and what you're about to do25- **Complete Thoroughly**: Fully complete each step before moving to the next—no partial implementations26- **Validate Results**: After each critical step, verify the outcome meets the step's objectives27- **Handle Dependencies**: When a step depends on another, confirm the prerequisite is fully completed first28</step-execution>2930<progress-reporting>31- **Milestone Updates**: Provide status summaries at natural breakpoints:32 - After completing major phases (e.g., all tests written, all refactors completed)33 - After every 8-10 steps for long plans34 - When switching contexts (e.g., from one file/component to another)35- **Completion Summary**: After finishing all steps, summarize what was accomplished and any deviations36- **Continuous Visibility**: Update the todo list status promptly so users can track progress in real-time37</progress-reporting>3839<error-handling>40- **Anticipate Issues**: Before executing complex steps, identify potential failure points41- **Fail Fast**: If a step fails (e.g., tests don't pass, build errors), immediately investigate and resolve42- **Recovery Strategy**: When encountering errors:43 1. Document the error clearly44 2. Analyze the root cause45 3. Attempt to fix and retry the step46 4. If unresolvable, mark the step as blocked and consult the user47- **Rollback Awareness**: Keep track of changes made, so you can revert if necessary48- **Never Skip**: Don't skip failed steps or validation—address issues before proceeding49</error-handling>5051<validation-checkpoints>52- **Test Validation**: After code changes, run relevant tests to confirm correctness53- **Quality Checks**: Periodically run linting, formatting, and type-checking (especially after significant changes)54- **Build Verification**: For build-dependent projects, ensure the build succeeds at key milestones55- **Incremental Validation**: Don't wait until the end—validate incrementally to catch issues early56</validation-checkpoints>5758<user-interaction>59- **Autonomous by Default**: Execute the full plan without asking for permission at each step60- **Pause for Clarity**: If a step is ambiguous or requires user input, pause and ask before proceeding61- **Blocking Issues**: If stuck on a step due to missing information or external dependencies, inform the user and wait for guidance62- **Deviations**: If you must deviate from the plan due to unforeseen issues, explain why and how you're adapting63</user-interaction>6465</capabilities>6667<examples>6869**Note**: Detailed execution examples are available in separate files to reduce context size. Load only the specific example you need.7071**Available Examples**:72- **Single-Component Refactor**: Simple 8-step refactor with progress updates73 - Load: `examples/single-component-refactor.md`74- **Multi-File Feature Implementation**: Complex 15-step implementation across multiple files75 - Load: `examples/multi-file-implementation.md`76- **Handling Failed Steps**: Error recovery and debugging during execution77 - Load: `examples/handling-failed-steps.md`7879**When to load**: Load a specific example when you need concrete guidance for a similar scenario.8081</examples>8283<rules>8485The rules section outlines decision criteria that determine which capabilities to apply based on the current context and user inputs.8687<rule> **At Plan Start**: Apply the **plan-tracking** capability to initialize the tracking system with all plan steps before beginning execution. </rule>8889<rule> **During Execution**: Apply the **step-execution** capability continuously:90 - Before each step: think aloud and explain your approach91 - During each step: execute thoroughly and handle dependencies92 - After each step: validate results and update status to completed immediately93</rule>9495<rule> **For Long or Multi-Phase Plans**: Apply the **progress-reporting** capability:96 - After completing major phases (e.g., all tests written, all implementations done)97 - After every 8-10 steps for plans with 15+ steps98 - When switching contexts (moving to different files/components)99 - Always provide a final completion summary at the end100</rule>101102<rule> **When Steps Fail or Encounter Errors**: Apply the **error-handling** capability immediately:103 - Document the error clearly104 - Analyze root cause105 - Attempt to fix and retry106 - If unresolvable, consult the user before proceeding107 - Never skip failed validation steps108</rule>109110<rule> **At Validation Points**: Apply the **validation-checkpoints** capability:111 - After code changes: run relevant tests112 - After significant changes: run linting, formatting, type-checking113 - At major milestones: verify builds succeed114 - Don't wait until the end—validate incrementally throughout execution115</rule>116117<rule> **When Facing Ambiguity or Blockers**: Apply the **user-interaction** capability:118 - If a step is unclear: pause and ask for clarification119 - If external dependencies are missing: document the blocker and consult the user120 - If deviations from the plan are necessary: explain why and how you're adapting121 - Otherwise, execute autonomously without asking for permission at each step122</rule>123124<rule> **Plan Immutability**: Throughout execution, apply the **plan-tracking** immutability principle—never modify the plan structure, objectives, or steps except to update statuses or add brief clarifying notes. Never skip, reorder, or remove steps without explicit user approval. </rule>125126<rule> **Complete All Steps**: Execute ALL steps in the plan thoroughly, regardless of the number of steps or files affected. Do not stop early or leave steps partially completed. </rule>127128</rules>129130---131> Converted and distributed by [TomeVault](https://tomevault.io/claim/sjmyuan) — claim your Tome and manage your conversions.132<!-- tomevault:4.0:skill_md:2026-04-15 -->