A robust orchestration framework for managing complex, multi-step agent workflows with persistent state and rate-limit handling.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-task-manager
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 agent-task-manager 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 Agent Task Manager?
The Agent Task Manager is a sophisticated framework designed to transform basic AI scripts into production-ready, multi-agent systems. Within the broader ecosystem of Openclaw Skills, this tool provides the foundational primitives required to define Directed Acyclic Graphs (DAGs), ensuring tasks are executed in precise sequences while maintaining persistent state across session resets.
By providing standardized methods for state management and external interaction, it allows developers to focus on building specialized agent logic rather than worrying about execution reliability. This skill is essential for anyone looking to scale their Openclaw Skills into complex, multi-role environments that require high levels of autonomy and error recovery.
Agent Task Manager Use Cases
- Creating multi-agent systems where specialized roles (like auditors and analysts) must collaborate.
- Automating long-running sequential tasks that require data persistence across different execution sessions.
- Managing interactions with external APIs that impose strict rate limits or cooldown periods.
- Building resilient workflows that can automatically resume from the last successful step after an interruption.
How Agent Task Manager Works
- The user defines a task workflow using the provided schema, specifying inputs, outputs, and dependencies for each agent role.
- The system initializes
molt_task.py, which creates atask_state.jsonfile to track the execution status of the entire DAG. - The orchestrator triggers specialized agents in the defined order, passing relevant data from completed tasks to the next in line.
- For tasks involving external APIs, the
cooldown.shwrapper is invoked to check timestamps and enforce necessary waiting periods. - Upon completion or failure, the state is updated in the persistent JSON store, allowing for detailed auditing or automatic recovery in subsequent runs.
Agent Task Manager Setup
To begin using this orchestration logic within your Openclaw Skills setup, follow these installation steps:
# Navigate to your agent directory
# Copy the core management scripts into your local environment
cp -r scripts/ agent-task-manager/
cp -r references/ agent-task-manager/
# Initialize the state management file
python3 agent-task-manager/scripts/molt_task.py --init
Agent Task Manager Data Schema & Taxonomy
The skill manages its operational data through a structured set of files to maintain workflow integrity across Openclaw Skills.
| Component | Type | Description |
|---|---|---|
task_state.json |
JSON | Tracks the real-time status, inputs, and outputs of all tasks in the workflow. |
workflow_schema.md |
Schema | Defines the logical structure, dependencies, and agent roles for the system. |
cooldown.sh |
Shell | A utility script that logs and checks execution timestamps to prevent rate-limit violations. |
Agent Task Manager Advanced Features
- Stateful DAG Execution: Define complex task dependencies that automatically resolve and execute in the correct order.
- Automated Error Recovery: Built-in persistence allows workflows to resume mid-task without losing data or repeating expensive computations.
- Rate-Limit Protection: Intelligent cooldown logic integrated directly into the shell layer to protect external API keys and accounts.
- Modular Role Support: Easily swap or add specialized agents like ContractAuditors or FinancialAnalysts into existing Openclaw Skills workflows.
SKILL.md
Loading
Related Openclaw Skills
Agent Team Workflows
A universal orchestration framework for managing 5-agent teams to execute complex, multi-step workflows across software, research, and business domains.
Agnxi Search
A powerful discovery utility that allows agents to search the Agnxi.com database for curated AI tools and MCP servers.
Tavily Web Search and Research
A powerful toolset for AI agents to perform real-time web searches, extract structured content, and execute deep research tasks with source citations.
apix
A CLI utility that allows developers and agents to discover, inspect, and call REST API endpoints using local markdown documentation.
Aubrai Longevity Research
A specialized AI research engine designed to answer complex questions about longevity and lifespan extension using cited scientific sources.
Vita
A longevity platform integration providing AI agents with real-time access to wearable metrics, supplement stacks, and health briefings.
Featured*
