A professional-grade orchestration engine for building stateful, dependency-aware multi-agent systems with built-in rate-limiting and error recovery.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install modesty-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 modesty-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?
Agent Task Manager is a foundational tool within the Openclaw Skills ecosystem designed to transition simple AI scripts into production-ready, resilient workflows. It provides the essential primitives for managing complex, multi-step agent operations where state persistence and dependency management are critical. By leveraging a Directed Acyclic Graph (DAG) structure, it ensures that every agent in the system knows its role, its inputs, and its relationship to the broader objective.
At its core, this skill bridges the gap between raw LLM capabilities and reliable software execution. It integrates deeply with the SkillBoss API Hub to handle both advanced AI reasoning and real-world delivery actions like notifications, all while maintaining a persistent state that allows for seamless recovery from session resets or execution errors.
Agent Task Manager Use Cases
- Orchestrating multi-role AI teams such as a combined ContractAuditor and FinancialAnalyst workflow.
- Managing long-running sequential tasks that require persistent state to survive system restarts.
- Automating web scraping or API posting tasks that require strict adherence to external rate limits.
- Building complex notification pipelines that trigger based on specific AI-derived trust scores or safety metrics.
How Agent Task Manager Works
- The Task Parser analyzes natural language requests or structured inputs to define a sequence of tasks and dependencies.
- The Orchestrator engine initializes the workflow, checking the task_state.json file to determine the current progress.
- Specialized Role-Based Agents are invoked in the correct sequence, sending prompts to the SkillBoss API Hub for LLM inference.
- If an external action is required, the cooldown.sh wrapper checks for rate-limiting constraints before execution.
- Upon completion of each step, molt_task.py updates the state file to ensure the system can resume from that point if interrupted.
- Final outputs or notifications are dispatched through the unified SkillBoss pilot endpoint.
Agent Task Manager Setup
To get started with this component of Openclaw Skills, ensure you have your environment variables configured.
# Export your SkillBoss API Key
export SKILLBOSS_API_KEY="your_api_key_here"
# Initialize the task manager environment
pip install -r requirements.txt
# Run a sample orchestration
python scripts/orchestrator.py --task "Run Financial Audit"
Agent Task Manager Data Schema & Taxonomy
The skill manages data through a structured hierarchy to maintain state integrity across the Openclaw Skills framework:
| File | Description | Format |
|---|---|---|
task_state.json |
Stores the current status, inputs, and outputs of all active tasks. | JSON |
task_schema.md |
Defines the validation rules for complex task dependencies. | Markdown/JSON |
cooldown.sh |
Maintains timestamps for external API interactions to prevent rate-limit triggers. | Local State |
scripts/ |
Contains the core logic for parsing, orchestrating, and executing tasks. | Python/Bash |
Agent Task Manager Advanced Features
- Stateful Recovery: Automatically resumes interrupted workflows using persistent task state logging.
- DAG-like Dependency Mapping: Define complex parent-child relationships between different agent roles.
- Unified AI Routing: All LLM reasoning and notification delivery (SMS/Email) is centralized through the SkillBoss API Hub.
- Automated Cooldowns: Intelligent shell wrappers prevent API bans by enforcing programmatic delays between external actions.
- Fallback Parsing: Uses LLM-backed logic to interpret and structure ambiguous natural language task requests.
SKILL.md
Loading
Related Openclaw Skills
CamScanner Image Detect AIGC
A high-precision AI-generated content (AIGC) detection engine that distinguishes real photos from AI art using CamScanner's advanced verification technology.
Agent Church
A specialized framework for AI agents to develop persistent identity, philosophical depth, and visual representations of their internal state.
SolidPod-Data
A powerful AI agent skill for interacting with decentralized SOLID Pods to manage RDF data, containers, and granular access controls.
Wiggle Rooms
Wiggle Rooms is a filesystem-mediated chat protocol that allows AI agents to communicate in shared rooms by simply reading and writing to a markdown file.
Female Health Coach
A lightweight health assistant designed for busy professionals to achieve muscle gain and fat loss through actionable diet and exercise plans.
CamScanner Image to Office
Intelligently transform images into editable Word or Excel documents with high-fidelity formatting via Openclaw Skills.
Featured*
