Skip to content

bhuvanprakash/agentguard

Repository files navigation

AgentGuard Logo

AgentGuard

The Protocol-Neutral Enforcement Proxy for AI Agents

Build Status Latest Release License Documentation


🛡️ Secure Your AI Agents in Production

AgentGuard is a high-performance security proxy that sits between your AI agent and the real world. Every tool call, task delegation, and payment transaction is intercepted and validated against your custom policy before execution.

✨ Key Features

  • ⚡ Blazing Fast: Built in Go, adding <1ms overhead to your agent pipeline.
  • 🔌 Protocol Agnostic: Native support for MCP (Model Context Protocol), A2A, and ATXP.
  • 📜 Policy-as-Code: Human-readable YAML policies to allow, block, or escalate sensitive actions.
  • 🙋 Human-in-the-Loop: Seamlessly pause risky tool calls for manual approval via Dashboard or Slack.
  • 💰 Spend Control: Set granular daily USD limits and token-bucket rate limits per agent.
  • 🔒 Zero Trust: HMAC-SHA256 request signing ensures agents can't spoof their identity.

🚀 Quickstart

1. Launch with Docker

docker run -p 7777:7777 \
  -e AGENTGUARD_UPSTREAM_URL=http://your-api:8080 \
  -v $(pwd)/policy.yaml:/app/policy.yaml \
  bhuvanprakash/agentguard:latest

2. Define Your Policy

version: "1"
default: block

agents:
  - id: billing-agent
    allow:
      - tool: read_invoice
    escalate:
      - tool: send_payment   # Requires human approval
    block:
      - tool: delete_record
    spend_limit_daily_usd: 50.00

🧠 How it Works

AgentGuard intercepts JSON-RPC and REST calls from your agents. It evaluates the "Tool/Call" intent against your rules:

  1. Identity Verification: Validates the X-Agent-HMAC signature.
  2. Safety Check: Intercepts "Irreversible Tools" (like drop_table).
  3. Policy Evaluation: Matches the agent ID and tool name.
  4. Enforcement:
    • ALLOW: Forwards directly to the upstream.
    • BLOCK: Returns a 403 Forbidden.
    • ESCALATE: Returns 202 Accepted and waits for approval.

🛠️ SDK Support

Language Package Install
Node.js agentguard npm install agentguard
Python agentguard pip install agentguard

📊 Dashboard & Monitoring

While the proxy is fully standalone, you can connect it to the Nascentist Dashboard for:

  • 📈 Real-time spend analytics
  • 📱 Mobile approval notifications
  • 🛠️ Drag-and-drop policy editor
  • 🔐 Managed agent secret storage

📜 License

Distributed under the Apache License 2.0. See LICENSE for more information.


Built by Bhuvan Prakash for nascentist.com

DocumentationReport BugRequest Feature

About

AgentGuard is a high-performance security proxy that sits between your AI agent and the real world. Every tool call, task delegation, and payment transaction is intercepted and validated against your custom policy before execution.

Topics

Resources

License

Contributing

Stars

Watchers

Forks

Packages

 
 
 

Contributors