Task Pilot is a robust execution framework that prevents Claude from losing track of complex tasks by using verified subtasks and strategic context anchors.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install task-pilot
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install task-pilot using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Task Pilot?
Task Pilot is a specialized framework designed for Openclaw Skills that solves the two primary failure points of long-running AI tasks: scope drift and context loss. By forcing a plan-first approach, it ensures that execution never wanders away from the original goal, even as conversations grow in length.
The skill introduces a disciplined protocol where Claude plants context anchors every 8-12 exchanges. These anchors serve as recovery points that survive context compaction, allowing the agent to resume work from the exact point of interruption without repeating completed steps or guessing the current state.
Task Pilot Use Cases
- Managing complex coding projects that require more than 3 distinct development steps.
- Long-form research or content creation spanning multiple interaction cycles.
- Critical workflows where an error in an early stage would cause a catastrophic failure in subsequent steps.
- Collaborative tasks where human-in-the-loop confirmation is required at specific milestones.
How Task Pilot Works
- Plan Initialization: Before any action is taken, the agent generates a structured Task Plan including goals, success criteria, and numbered steps.
- Verification Gate: The agent waits for user confirmation of the plan to ensure alignment on scope.
- Step Execution & Checkpoints: After each step, the agent marks it as complete, verifies the output, and explicitly states the next action.
- Context Anchoring: Strategic snapshots are recorded periodically to preserve the working state against LLM context window limits.
- Completion Verification: The task is only finalized once all steps are checked and side effects (like file changes) are verified.
Task Pilot Setup
To get started with Task Pilot and the integrated ThinkStack, run the following commands in your terminal:
openclaw install task-pilot
# Highly recommended to install the full stack
openclaw install clarity-first
openclaw install thinkdeep
Task Pilot Data Schema & Taxonomy
Task Pilot utilizes a structured metadata taxonomy to track progression and maintain state stability. The data is organized into three primary blocks:
| Block Type | Contents | Purpose |
|---|---|---|
| Task Plan | Goal, Steps [ ], Success Criteria, Risks | Establishes the initial roadmap and scope boundaries. |
| Context Anchor | Completed Steps, Current Step, Key Decisions | Provides a recovery point for long-running sessions. |
| Completion Gate | Step Checklist, Result Summary, Verified Side-effects | Ensures all requirements are met before closing the task. |
Task Pilot Advanced Features
- Context Loss Recovery Protocol: Automatically identifies the most recent anchor to resume work after context compaction.
- Scope Discipline Enforcement: Prevents 'feature creep' by requiring a plan update for any requirements added mid-task.
- ThinkStack Integration: Works seamlessly with clarity-first for requirement gathering and thinkdeep for complex decision-making.
- Verification Loops: Explicitly flags failures and prevents silent progression to subsequent subtasks.
SKILL.md
Loading
Related Openclaw Skills
Blog Watcher
Blog Watcher is an automated monitoring tool that tracks website updates and RSS feeds to keep you informed of new content instantly.
Thinkdeep
Thinkdeep is a structured reasoning protocol that forces AI agents to use step-by-step analysis and self-critique to eliminate hallucinations and improve accuracy.
Spotify Controller
A terminal-driven Spotify integration that enables full playback control, track searching, and playlist management through a CLI interface.
Health Tracker
A personal health and activity monitoring system designed to track hydration, sleep cycles, and daily step counts using local JSON storage.
RSS Monitor
A streamlined automation skill for monitoring RSS and Atom feeds to detect and extract new content updates programmatically.
Clarity First
Clarity First is an intent detection protocol that eliminates wasted AI work by surfacing hidden assumptions and clarifying requirements before execution starts.
Featured*
