Clawlist 🦞

A task management system for OpenClaw agents. Plan, execute, track, and verify tasks with checkpoint validation.

Overview

Clawlist provides a structured workflow for any multi-step project, long-running task, or infinite monitoring workflow. It breaks work into manageable phases with clear verification at each step.

The 4-Phase Workflow

┌─────────────┐    ┌─────────────┐    ┌─────────────────────────┐    ┌─────────────┐
│ Brainstorm  │ →  │  Write Plan │ →  │ Doing-tasks / Dispatch  │ →  │ Verify Task │
│  (clarify)  │    │  ( roadmap) │    │      (execute)          │    │  (confirm)  │
└─────────────┘    └─────────────┘    └─────────────────────────┘    └─────────────┘

Phase 1: Brainstorm

Ask questions, explore approaches, validate understanding before building.

Phase 2: Write Plan

Create detailed implementation plans with:

• Bite-sized tasks (2-5 min each)
• Checkpoints with verification criteria
• Execution options (single-agent or parallel)

Phase 3: Execute

Choose your execution mode:

Mode	Use When	Command
Single-Agent	Sequential tasks, dependencies	doing-tasks
Parallel	2+ independent tasks	dispatch-multiple-agents

Phase 4: Verify

Confirm completion against the original plan. Always run this after execution.

Installation

Via ClawdHub (Recommended)

clawdhub install clawlist

Manual Install

# Clone to your workspace skills directory
git clone https://github.com/arisylafeta/clawlist.git ~/.openclaw/workspace/skills/clawlist

Usage

Basic Project Flow

# 1. Brainstorm - clarify what you're building
Skill: brainstorming

# 2. Write plan - create roadmap  
Skill: write-plan

# 3. Execute - do the work (choose one)
Skill: doing-tasks          # Sequential
# OR
Skill: dispatch-multiple-agents  # Parallel

# 4. Verify - confirm completion
Skill: verify-task

Example: Building a Feature

Input: "Build me a todo app"

Agent flow:

1. Brainstorm: "What features? Web or mobile? Storage?"
2. Write Plan: Creates checkpoints for setup, UI, backend, testing
3. Execute: Works through tasks sequentially
4. Verify: Confirms all features work, tests pass

Example: Parallel Research

Input: "Research 5 competitors"

Agent flow:

1. Brainstorm: Define what to research (features, pricing, etc.)
2. Write Plan: Creates 5 parallel research tasks
3. Dispatch: Spawns 5 agents simultaneously
4. Verify: Integrates findings, confirms coverage

Task Types Supported

Type	Duration	Example
One-off	Minutes-hours	Fix a bug, send message
Project	Days-weeks	Build feature, write report
Long-running	Ongoing	GitHub issue monitoring
Infinite	Forever	Moltbook engagement, research

Task Tracking

Clawlist integrates with memory/tasks/ongoing-tasks.md for heartbeat monitoring:

• Tracks task health (🟢🟡🔴)
• Detects blockers automatically
• Pings user when intervention needed
• Archives completed tasks

Skills Included

Skill	Purpose
brainstorming	Clarify intent, explore approaches
write-plan	Create detailed implementation plans
doing-tasks	Sequential task execution
dispatch-multiple-agents	Parallel task execution
verify-task	Confirm completion, document outcomes

Configuration

Add to your HEARTBEAT.md:

## Ongoing Tasks Check (every check)
1. Read memory/tasks/ongoing-tasks.md
2. For each active task:
   - Check if last run was successful
   - If blocked >1h: ping user
   - If failed: alert user immediately
   - Update health status (🟢🟡🔴)
3. Execute any due infinite tasks
4. Update ongoing-tasks.md with current status

Best Practices

1. Always brainstorm first — Don't jump into execution without clarity
2. Write plans before doing — The plan is your roadmap and verification criteria
3. Choose execution mode wisely — Parallel for independent tasks, sequential for dependencies
4. Always verify — Confirm completion against the plan, not assumptions
5. Track in ongoing-tasks.md — For long-running/infinite tasks

Links

• ClawdHub: https://www.clawhub.ai/arisylafeta/clawlist
• Repository: https://github.com/arisylafeta/clawlist

License

MIT

---

Built with 🦞 for the OpenClaw ecosystem