Runtime policy enforcement
for AI agents.

Detect dangerous misconfigurations before they become exposed attack surfaces.

Join the Waitlist

The Problem

AI agents are deployed with dangerous defaults. Every day.

Publicly bound to 0.0.0.0 with no authentication

Running inside privileged or root containers

API keys sitting in plaintext .env files

Config files world-writable on shared hosts

These are not theoretical risks — they are common misconfigurations that ClawShield detects today.

What is PolicyGate?

A runtime policy enforcement framework for AI agents.

PolicyGate provides modular, rule-based security tools that audit agent deployments against configurable policy sets. Each module targets a specific attack surface — network exposure, container posture, secrets handling, file permissions — and outputs deterministic, automation-ready results.

PolicyGate is designed for CI pipelines, pre-deploy checks, and continuous monitoring. It exits non-zero when security thresholds are exceeded.

ClawShield

The first PolicyGate module. Available now.

ClawShield v0.3.0

Security audit tool for OpenClaw deployments

What it checks

  • Public bind address with auth disabled
  • Containers running as root or privileged
  • API keys in .env and config files
  • World-writable config and secrets files

Install

pip install clawshield

Run

clawshield path/to/openclaw.yaml

CI Mode

clawshield --fail-on high --json config.yaml

How It Works

1

Scan

Modular scanners collect facts from runtime configs, containers, environment files, and file permissions.

2

Evaluate

The policy engine evaluates collected facts against YAML-defined rules with configurable severity levels.

3

Report

Structured JSON output with deterministic schema — designed for CI pipelines and automation.

Roadmap

What's coming next.

Continuous Monitoring

Watch mode for live detection of configuration drift.

Additional Runtime Adapters

Support for more AI agent frameworks beyond OpenClaw.

Policy Bundles

Pre-built policy sets for common deployment profiles.

Expanded Docker Checks

Deeper container hardening analysis and network policies.

Advanced Secrets Detection

Entropy-based analysis and broader credential formats.

Signed Policy Packs

Tamper-proof, verifiable policy distribution.

Get Early Access

Join the waitlist for PolicyGate updates, new modules, and early access.

No spam. Unsubscribe anytime.