Skip to content

tim-dickey/OpenClaw-agents

OpenClaw Agents Project

License: MIT Contributions Welcome

A community collection of downloadable, editable agent and sub-agent personas for use with OpenClaw

OpenClaw Agent Project


What is OpenClaw?

OpenClaw is an open-source personal AI assistant that runs on your own machine. It connects to any chat app — WhatsApp, Telegram, Discord, Slack, Signal, iMessage — and supports persistent memory, browser control, skills/plugins, and full system access. You own your AI, your data, and your agents.


What Are Agents and Sub-Agents?

Agents are fully defined AI personas that you load into OpenClaw. Each agent has a unique name, personality, communication style, set of capabilities, and behavioral rules. Think of an agent as a dedicated team member with a specific role.

Sub-agents are focused, task-specific modules that a parent agent can delegate to. For example, an executive assistant agent might delegate email triage to an email-manager sub-agent, and research tasks to a web-researcher sub-agent. Sub-agents are designed to be composable and reusable.


How to Use

  1. Clone this repository (or browse it on GitHub)
  2. Browse the agents/ or sub-agents/ directories to find a persona that fits your needs
  3. Copy agent.md into your OpenClaw agents directory (for sub-agents, place sub-agent.md under sub-agents/)
  4. Customize the file to match your preferences, tone, and context
  5. Restart or hot-reload your OpenClaw instance
  6. Engage via your chosen chat channel — WhatsApp, Telegram, Discord, or others

For detailed setup instructions, see docs/getting-started.md. For customization help, see docs/customization-guide.md. For executable behavioral contract tests, see tests/README.md.


Quality Signals

This repository now uses two complementary validation layers for agent behavior:

  • evals/ holds deterministic offline fixtures with schema and target-reference validation only
  • tests/ holds executable pytest-based behavioral contract tests for TDD workflows

Use fixtures to document intended behavior in a reviewable format, and use behavioral contract tests when you want runnable pass/fail checks.


Table of Contents


Featured Agents

Agent Category Description
Aria Personal Morning briefing agent — calendar, weather, news, tasks
Maxwell Professional Executive assistant — email triage, scheduling, follow-ups
Hex Developer Code reviewer — bugs, style, security, GitHub integration
Lyra Creative Content writer — blog posts, social media, newsletters
Velocity Team Scrum Master | Coach — Daily Scrum facilitation, Sprint Backlog tracking, impediment alerts

Featured Sub-Agents

Agents declare required sub-agent installs in frontmatter sub_agents using sub-agent folder slugs. If an agent README mentions a dependency, browse the matching category under sub-agents/ to find it.

Sub-Agent Category Description
web-researcher Research Targeted web research with structured summaries
email-manager Communication Email triage and draft replies
task-tracker Productivity Task capture, categorization, and updates
data-analyst Data Data analysis with insights and anomaly flags

Contributing

We welcome contributions from the OpenClaw community! To submit a new agent or sub-agent:

  1. Use the new agent issue template to propose your idea
  2. Fork this repo and create your agent using the template
  3. Open a pull request — your submission will be automatically validated by CI

See CONTRIBUTING.md for full guidelines, quality standards, and naming conventions.


License

This project is licensed under the MIT License. All contributed agents and sub-agents are shared under the same license unless otherwise noted.

About

Community-driven repository of reusable OpenClaw agents and sub-agents with validation workflows, fixtures, and tests.

Topics

Resources

License

Code of conduct

Contributing

Security policy

Stars

Watchers

Forks

Packages

 
 
 

Contributors