AgentVeil: security tool for Hermes Agent

AgentVeil

agentveil-protocol/agentveil-sdk

Policy layer that allows, gates or redirects risky AI agent actions on configured routes

In short

AgentVeil is an agent security tool that checks supported actions and applies policy before they run. It lists Hermes CLI among the tools it works with.

What AgentVeil does

On configured routes, AgentVeil checks supported actions and applies a policy: allow, request approval, provide redirect guidance to a controlled path, or block. The pitch is to stop risky actions without stopping the work, so a safe alternative is offered when a controlled path exists.

It is distributed as agentveil-mcp-proxy and agentveil packages on PyPI, with a console at agentveil.dev. The README describes a loop of connectors, bounded evidence, scope and design basis sections, and lists compatibility with Cursor, Claude Code, Codex, Gemini CLI, Hermes CLI and MCP tools.

The opening part of the README is short on setup specifics beyond the pip command, so confirm the Hermes connector details in the full documentation. Policy applies only to supported actions on routes that have been configured.

Key features

Policies to allow, request approval, redirect or block actions
Safe-alternative guidance when a controlled path is available
MCP proxy package on PyPI
Web console
Works with Cursor, Claude Code, Codex, Gemini CLI, Hermes CLI and MCP tools

When to use it

Requiring approval before an agent runs a risky action
Putting a policy proxy in front of MCP tools
Auditing what agents attempted on configured routes

Who it is for: Teams that want approval and policy control over what their AI agents are allowed to do.

How it fits with Hermes Agent

It is a general agent security tool that names Hermes CLI among the clients it works with.

How to install AgentVeil

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

pip install agentveil-mcp-proxy

FAQ

What is AgentVeil?

AgentVeil is an independent security layer for AI agents. On configured routes it allows, gates, redirects or blocks supported actions.

Does AgentVeil work with Hermes Agent?

The README lists Hermes CLI among the tools it works with, alongside Cursor, Claude Code, Codex and Gemini CLI. The excerpt gives no Hermes-specific steps.

How do I install AgentVeil?

Run pip install agentveil-mcp-proxy. A separate agentveil package is also published on PyPI.

Is AgentVeil free and open source?

The GitHub license field says MIT, though the README links to a LICENSING.md file. Read that file before reuse.