A self-supervising manager for long-running AI agent tasks with progress tracking and periodic reporting.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install task-supervisor
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-supervisor 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 Supervisor?
Task Supervisor is a specialized skill designed to handle complex, multi-step operations that require extended execution times. By transforming standard agent interactions into managed workflows, it ensures that high-latency tasks—such as multi-domain research, large-scale deployments, or overnight processing—are handled reliably without losing progress. This skill is a core component of the Openclaw Skills ecosystem, providing the autonomy needed for agents to function as independent workers.
At its core, Task Supervisor focuses on state persistence and transparency. It creates structured documentation for every large task, decomposing goals into actionable steps and maintaining a real-time log of findings and errors. This allows the AI agent to recover from failures, resume from specific checkpoints, and keep users informed throughout the lifecycle of the task without requiring constant manual oversight.
Task Supervisor Use Cases
- Managing complex software deployments that require more than 20 minutes of runtime or 5+ distinct steps.
- Executing "overnight" tasks where the user needs a summary delivered to their messaging app the next morning.
- Coordinating cross-domain workflows such as simultaneous web research, technical writing, and cloud deployment.
- Implementing persistent state for multi-agent workflows that use sub-agents to complete modular objectives.
- Tracking progress on research projects involving the synthesis of multiple large documents or datasets.
How Task Supervisor Works
- Identification: The agent evaluates incoming requests to determine if they qualify as "Large" based on time, step count, or specific user triggers like "take your time."
- Task Initialization: A dedicated markdown file is created in the .tasks/ directory using a kebab-case slug for persistent tracking.
- Decomposition: The high-level objective is broken down into a numbered list of steps within the task file.
- Cron Spawning: A background cron job is initialized via the CLI to trigger periodic status reports to the user.
- Execution & Update: The agent executes each step sequentially, updating the markdown checkboxes and logs with timestamps and key findings immediately after each action.
- Reporting: At set intervals, the agent reads the task file and sends a concise summary to the user via integrated communication channels.
- Completion: Once all steps are finished, the agent generates a final result summary, sets the status to done, and removes the reporting cron job.
Task Supervisor Setup
To integrate Task Supervisor into your workflow, ensure your environment supports directory management and cron scheduling. The skill operates within the Openclaw Skills framework and uses standard CLI commands for automation. To manually initialize a reporting interval for a long-running task, you can use the following bash command:
openclaw cron add "task-report-<SLUG>" \
--schedule "*/15 * * * *" \
--message "Read .tasks/<SLUG>.md and send a progress update. Include: completed steps, current step, and any blockers." \
--once-complete
Ensure that your agent has read/write permissions for the .tasks/ folder in your project root.
Task Supervisor Data Schema & Taxonomy
The Task Supervisor skill organizes data using a clean, markdown-based taxonomy for maximum human and machine readability:
| Component | Detail |
|---|---|
| Storage Path | .tasks/<task-slug>.md |
| File Format | Markdown (GFM) with Status, Started, and Last Updated metadata |
| Status States | in_progress, paused, done, failed |
| Step Tracking | GFM Checkboxes: [ ] (pending), [x] (done), [!] (failed) |
| Log Structure | Chronological entries including ISO timestamps and technical summaries |
| Metadata | Includes estimated step count and final result summaries |
Task Supervisor Advanced Features
- Intelligent Resume: Automatically detects the last completed step in a
.tasks/file to continue interrupted work without repeating steps. - Dynamic Interval Reporting: Adjusts the frequency of Openclaw Skills cron reports based on task duration (e.g., 10 min for quick tasks, 30 min for long tasks).
- Multi-Task Heartbeats: Maintains awareness of multiple concurrent tasks and provides a consolidated status during agent heartbeat responses.
- Error Recovery Protocols: Specifically marks failed steps with the
[!]flag and attempts alternative approaches before pausing the workflow. - External Integration: Supports immediate alerts via Feishu or WhatsApp when a task becomes "stuck" or requires manual user intervention.
SKILL.md
Loading
Related Openclaw Skills
Auto-Memory-Distiller
A background service that distills raw OpenClaw conversation logs into structured, theme-based, and searchable Markdown memory cards.
Cloudflare DNS Management Skill
A powerful tool for programmatically managing Cloudflare DNS records to automate domain configuration and email authentication.
Time Clawshine
An automated, restic-powered backup system that provides hourly incremental snapshots for your OpenClaw instance to prevent data loss or agent errors.
Bible Research
A specialized AI agent for in-depth Bible study and scriptural research using a structured, context-aware methodology.
Clawd Modifier
A creative customization toolkit for personalizing the Claude Code mascot's appearance and color schemes within your terminal.
Abstract Onboard
A comprehensive toolkit for AI agents to deploy contracts, bridge assets, and interact with the Abstract ZK Stack Layer 2 network.
Featured*
