Our review
Executes a predefined skill sequence from task_plan.md step by step, updating progress and handling errors deterministically.
Strengths
- Deterministic and reproducible execution of a skill sequence.
- Integrates error handling with recovery protocols.
- Automatically updates progress and findings.
- Tracked via beads tasks for each step.
Limitations
- Requires a prior plan created by continuous-planner.
- Depends on the exact format of task_plan.md.
- Failures may require user intervention.
Use after creating a continuous plan to automatically execute the skill sequence.
Do not use if you have no pre-existing plan or if the steps are simple enough to execute manually.
Security analysis
CautionThe skill is an orchestrator that reads a plan file and executes skills sequentially using allowed tools including Bash. While the skill itself contains no malicious code, it facilitates the execution of arbitrary commands and skills defined in the task plan, inheriting risk from its inputs. No obfuscation or direct destructive instructions are present.
- •The skill uses Bash and Skill tools, which can execute arbitrary commands and invoke other skills, potentially leading to destructive actions if the task plan includes malicious steps.
Examples
/run-plan/run-plan .claude/continuous/task_plan.mdname: run-plan description: "Execute a continuous plan deterministically by following the skill sequence. Use after continuous-planner creates a plan with skill sequences. Reads task_plan.md and executes each skill step by step." disable-model-invocation: true allowed-tools: Read, Write, Edit, Bash, Skill
Run Plan - Deterministic Skill Execution
Execute continuous plans by following the skill sequence step-by-step.
When To Use
After creating a continuous plan with continuous-planner, run this to execute the skill sequence deterministically.
How It Works
- Read task_plan.md to get the skill sequence
- For each skill in sequence:
- Check if status is "⏸️ Pending"
- If pending, execute the skill
- Update status in task_plan.md
- Update progress.md with results
- Handle failures with recovery protocols
- Mark plan complete when all skills executed
Skill Sequence Format
The task_plan.md should have a Skill Sequence table:
| Step | Skill | Purpose | Input | Output | Status |
|------|-------|---------|-------|--------|--------|
| 1 | front-door | Interview/triage | User request | requirements | ⏸️ Pending |
| 2 | api-designer | Design API | requirements | api_spec | ⏸️ Pending |
| 3 | implement-plan | Execute | api_spec | Complete | ⏸️ Pending |
Execution Flow
┌─────────────────────────────────────────┐
│ 1. Read task_plan.md │
│ 2. Parse Skill Sequence table │
│ 3. For each pending skill: │
│ a. Invoke skill with input │
│ b. Capture output │
│ c. Update findings.md │
│ d. Update progress.md │
│ e. Update skill status in plan │
│ 4. Update beads │
│ 5. Report completion │
└─────────────────────────────────────────┘
Error Handling
If a skill fails:
- Log error to findings.md
- Check failure-recovery skill for protocols
- Update skill status to "❌ Failed"
- Ask user for guidance
Beads Integration
Each skill execution is tracked as a bead task:
- Plan epic:
continuous-[project-name] - Skill tasks: Created as subtasks
- Completion: Mark beads as complete
Example Usage
# From project directory with continuous plan
/run-plan
# Or specify plan location
/run-plan .claude/continuous/task_plan.md
Output
After execution:
- task_plan.md: All skills marked complete or failed
- findings.md: Research, errors, discoveries
- progress.md: Execution log with checkpoints
- beads: Tasks marked complete
Keywords
run plan, execute plan, deterministic, skill sequence, continuous plan, task_plan.md
Next.js App Router Expert
Development
A skill that turns Claude into a Next.js App Router expert.
README Generator
Development
Creates professional and comprehensive README.md files for your projects.
API Documentation Writer
Development
Generates comprehensive API documentation in OpenAPI/Swagger format.