agent-task-manager
Manages and orchestrates multi-step, stateful agent workflows; handles task dependencies, persistent state, error recovery, and external rate-limiting. Use for creating new multi-agent systems, improving sequential workflows, or managing time-bound actions.
Why use this skill?
Orchestrate complex, stateful multi-agent workflows with OpenClaw. Features built-in task dependency management, error recovery, and external API rate-limiting.
Install via CLI (Recommended)
clawhub install openclaw/skills/skills/dobbybud/agent-task-managerWhat This Skill Does
The Agent Task Manager skill serves as the foundational orchestration engine for the OpenClaw ecosystem, enabling developers to build sophisticated, multi-step agent workflows. Unlike standard linear scripts, this skill introduces a robust state-management system via molt_task.py, which tracks task dependencies in a Directed Acyclic Graph (DAG) structure. By storing execution state in task_state.json, it ensures that complex processes are atomic, resumable, and fault-tolerant. If an agent session is interrupted, the manager recognizes the last successful state, preventing redundant resource consumption. Additionally, the skill includes built-in rate-limiting primitives through cooldown.sh, ensuring that your agents can interact with sensitive external APIs (like social platforms or financial data providers) without triggering anti-bot mechanisms or violating usage quotas.
Installation
To integrate the Agent Task Manager into your OpenClaw environment, ensure you have the core framework installed. Execute the following command in your terminal:
clawhub install openclaw/skills/skills/dobbybud/agent-task-manager
Once installed, verify the installation by checking the scripts/ directory for the presence of molt_task.py and cooldown.sh.
Use Cases
- Multi-Agent Orchestration: Coordinate specialized agents (e.g., a Financial Analyst and a Contract Auditor) to work in sequence where the output of one serves as the mandatory input for the next.
- Rate-Limited API Integration: Automatically throttle tasks that interact with strict third-party APIs, ensuring compliance with global rate limits.
- Long-Running Audits: Perform complex, stateful tasks like scanning multiple smart contract addresses, where the agent needs to track which addresses were audited in case of system crashes.
- Sequential Decision Making: Chain tasks where each step depends on the validation results of the previous operation.
Example Prompts
- "Initialize a new task workflow named 'ProtocolAuditor' that first runs the ContractAuditor module, then passes the results to the FinancialAnalyst, and finally posts the summary to Moltbook."
- "Resume the pending task 'MoltFinance-Auditor' from the last state saved in task_state.json."
- "Configure a rate-limited sequence for web scraping that ensures a 60-second cooldown between each URL request using the cooldown wrapper."
Tips & Limitations
When designing your workflows, strictly define the inputs and outputs for each module to prevent data mismatches during handover. The task state relies on local file persistence, so ensure your OpenClaw container has appropriate write permissions for the task_state.json file. While the skill handles basic error recovery, you should still implement custom try-catch logic within your specialized agent roles. Avoid creating circular dependencies in your workflow schemas, as the DAG structure does not support cycles.
Metadata
Not sure this is the right skill?
Describe what you want to build — we'll match you to the best skill from 16,000+ options.
Find the right skillPaste this into your clawhub.json to enable this plugin.
{
"plugins": {
"official-dobbybud-agent-task-manager": {
"enabled": true,
"auto_update": true
}
}
}Tags(AI)
Flags: file-write, file-read, external-api, code-execution