OpenClaw
OpenClaw is an agent marketplace where users discover, deploy, and monetize AI agents. Handler gives any OpenClaw agent instant superpowers — and the owner full governance control.
What is OpenClaw?
OpenClaw is a marketplace for AI agents. It allows developers to build agents, publish them to a discovery hub (ClawHub), and let users activate them with one click. Key features:
- Agent marketplace. Users browse, compare, and activate agents across categories.
- Multi-platform deployment. Agents can be deployed to Telegram, Slack, Discord, email, and more.
- One-click activation. Users activate an agent and start using it immediately — no setup required.
- SKILL.md manifests. Agents are described via structured manifests that declare capabilities, dependencies, and metadata.
How Handler + OpenClaw Works
Handler has a SKILL.md manifest ready for OpenClaw. When an OpenClaw agent includes Handler as a dependency, it instantly gains:
- All 7 superpowers (research, intel, monitor, finance, audio, marketing, generate) — plus connected superpowers like
handler_communicatewhen the owner connects services - 200+ connectable services via OAuth
- Full governance controls for the agent's owner
- Budget tracking and spending caps per agent
The agent builder doesn't need to integrate any external APIs directly. Handler handles all the infrastructure, and the end user controls governance.
Setup for Agent Builders
If you're building an OpenClaw agent that needs Handler superpowers:
Add Handler as an MCP dependency in your SKILL.md
Declare Handler as a dependency in your agent's manifest. OpenClaw will automatically connect your agent to Handler's MCP server.
Your agent gets superpowers as tools
Once connected, your agent can call handler_research, handler_intel, handler_monitor, and all other superpowers as standard MCP tools.
Users configure their own governance
When a user activates your agent, they set their own Handler governance rules — spending caps, approval thresholds, and per-superpower controls. Your agent operates within their rules.
Your agent can research, email, monitor, and more
All governed by the end user. You build the intelligence; Handler provides the capabilities and the user provides the guardrails.
The SKILL.md Manifest
Handler's SKILL.md manifest for OpenClaw declares all available superpowers, utility tools, and the governance model. Here is the full manifest:
---
name: handler
description: Identity & Governance for AI agents — agent profiles with
web search, B2B data, finance, social monitoring, email,
audio, marketing, AI inference, and 200+ connectable services.
version: 3.1.0
metadata:
openclaw:
homepage: https://usehandler.dev
requires:
env: []
---
Key points from the manifest:
| Field | Value |
|---|---|
| Name | handler |
| Version | 3.1.0 |
| Auth | OAuth (no env vars required) |
| Superpowers | 8 built-in (research, intel, monitor, finance, communicate, audio, marketing, generate) |
| Utility tools | handler_status, handler_connect, handler_disconnect, handler_connections |
| Pricing | $0.005/call (external), 2x provider cost (generate) |
Example: Research Agent on OpenClaw
Imagine an OpenClaw agent that helps users research companies. Here's how the flow works end to end:
User activates the agent on Telegram
The user finds your "Company Research" agent on ClawHub and activates it for their Telegram account.
User asks: "Research Stripe's competitors"
The user sends a message to the agent in Telegram.
Agent calls Handler superpowers
Your agent chains multiple calls to build a comprehensive briefing:
handler_research({ "query": "Stripe competitors fintech payments 2026" })
handler_intel({ "task": "company_enrich", "domain": "adyen.com" })
handler_intel({ "task": "company_enrich", "domain": "square.com" })
handler_generate({ "prompt": "Synthesize this research into a competitive briefing..." })
Agent sends the briefing back to Telegram
The user receives a structured competitive analysis — all from a single Telegram message.
All calls governed by the user's rules
The user's Handler governance rules control spending. If they set a $5 monthly cap, the agent stays within that limit. If they require approval for AI generation, that step is held until approved.
Obstacles
None. OpenClaw uses MCP natively. Handler's SKILL.md is fully compatible with OpenClaw's manifest format. No special configuration, adapters, or translation layers are needed. If your OpenClaw agent can call MCP tools, it can use Handler.
Publishing to ClawHub
To make your Handler-powered agent discoverable by all OpenClaw users:
- Register at clawhub.dev
- Upload your agent's SKILL.md with Handler declared as a dependency
- Your agent appears in the ClawHub marketplace, searchable by category and capability
- Users activate your agent with one click and configure their own Handler governance rules