AgentMFA is an opt-in multi-factor authentication and human approval system that allows AI agents to request biometric verification before performing sensitive actions.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agentmfa
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 agentmfa 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 AgentMFA?
AgentMFA is a sophisticated security layer designed for the modern AI workforce. It operates as an opt-in approval system where the AI agent explicitly requests human intervention before executing sensitive or irreversible tasks. By integrating with MCP-compatible clients, it enables a seamless workflow where agents pause for authorization, and human operators receive push notifications to approve or reject actions using biometrics.
Built to enhance the reliability of autonomous tools, AgentMFA ensures that critical operations—such as infrastructure changes or production deployments—remain under human control. It provides a robust framework for managing trust between developers and AI agents, making it a cornerstone for those building secure workflows with Openclaw Skills.
AgentMFA Use Cases
- Authorizing production database schema changes or deletions
- Approving force-pushes or history rewrites in Git repositories
- Validating Kubernetes resource deletions or workload modifications
- Confirming Terraform apply or destroy commands in live environments
- Verifying financial transactions or bulk email dispatching
- Securing modifications to cloud infrastructure and DNS settings
How AgentMFA Works
- Registration: The agent calls
register_agentto establish its identity and ensure it is recognized by the system. - Request Initiation: When a risky action is detected, the agent invokes
request_approvalwith a detailed context of the proposed task. - Human Notification: The human operator receives a push notification on their mobile device containing the action details.
- Biometric Verification: The operator reviews the request and provides approval or rejection via biometrics.
- Wait & Resolve: The agent uses
wait_for_approvalto block until a decision is made, then proceeds or aborts based on the cryptographic proof received.
AgentMFA Setup
To get started with this skill, you must install the CLI and authenticate your session:
# Install the AgentMFA CLI using Homebrew
brew install agentmfa/cli/agentmfa
# Authenticate and log in via your browser
agentmfa auth login
Once authenticated, ensure your MCP-compatible client is configured to use the agentmfa serve command.
AgentMFA Data Schema & Taxonomy
The skill manages identity and approval state through the following data structures:
| Data Point | Description |
|---|---|
| Agent Identity | Includes repository, branch, machine name, and code signature. |
| Request Object | Contains action (short label), description, and context (full details). |
| Response Metadata | Includes approved status, request_id, approved_by, and server_time. |
| Security Tokens | One-time proofs of approval and TOTP verification data. |
AgentMFA Advanced Features
- Multi-Agent Support: Seamlessly manage different agent roles and identities across various repositories.
- Biometric Integration: Leverages mobile device security for high-fidelity human-in-the-loop verification.
- Non-blocking Polls: Use
check_approval_statusfor sophisticated async workflows without blocking agent execution. - Identity Auditing: Detailed local identity detection via
agent_infoto track machine and branch context for every request. - Custom Service Scoping: Define specific services within approval requests for granular security control.
SKILL.md
Loading
Related Openclaw Skills
skylv-dependency-updater
A smart dependency management tool that scans, analyzes risk, and updates project packages across multiple languages.
Rate Limit Manager
An intelligent management skill designed to control API request rates and enforce usage quotas across multiple services for AI agents.
AI Prompt Optimizer
An automated prompt engineering tool that optimizes AI instructions through A/B testing, data-driven analysis, and token compression.
Content Brief for Hire
A professional-grade tool for generating clear, actionable content briefs that minimize friction when working with freelance writers, designers, and agencies.
Deloitte Insights & Corporate Intelligence
A comprehensive intelligence skill for analyzing Deloitte's global professional services, revenue models, and industry-leading consulting practices.
CRISIL India Ratings Research
A comprehensive research skill for analyzing CRISIL, India's dominant credit rating agency and a key S&P Global subsidiary.
Featured*
