OPEN SOURCE · APACHE 2.0

AI agents are overprivileged
by default.

Secure every agent action — from install to runtime. In code, not prompts.

View on GitHub
26%
of 31K skills have vulns
230+
malicious skills on ClawHub
140K+
OpenClaw stars, 0 governance
The Problem

Every tool scans. Nothing enforces.

Cisco, Snyk, and Caterpillar all scan before installation — then walk away. OpenClaw has 140K stars and gives agents full computer control with zero runtime governance. Prompt-based guardrails can be bypassed by prompt injection. You need enforcement in code, outside the LLM context window entirely.

Existing tools

Static scanners that check at install time, then stop watching. No runtime enforcement. No audit trail. No skill chaining controls. Lateral data movement between tools goes undetected. You're left hoping your prompts hold.

AgentWard

"Telling an agent 'don't touch the stove' is a natural-language guardrail that can be circumvented. AgentWard puts a physical lock on the stove — code-level enforcement that prompt injection can't override."

Security Lifecycle

From zero to fully secured — in one pipeline.

Every enterprise security workflow maps to five CLI commands. Run them individually or let agentward init wire the whole chain in one shot.

01 · SCAN
Detect
agentward scan

Catches load-time attack vectors — unsafe YAML constructs, pickle/marshal deserialization, and executable hooks that fire before runtime protection can help. Built to flag ClawHavoc-style supply chain attacks embedded in skill metadata.

→ Risk report
02 · CONFIGURE
Policy
agentward init

Auto-generates a suggested policy YAML from scan findings. Review and tune resource permissions, chaining rules, approval gates, and data classification boundaries.

→ agentward.yaml
03 · ENFORCE
Intercept
agentward inspect

Runtime proxy intercepts every tool call. Blocks unauthorized access, redacts PII, enforces skill-chain depth limits, and requires human approval for sensitive operations.

→ Live proxy
04 · VERIFY
Probe
agentward probe

Fires adversarial probes through the live proxy to prove policy catches what the scanner flagged. CI-ready with exit codes — fails the build if enforcement gaps exist.

→ Probe results
05 · MONITOR
Audit
audit logs

JSONL + RFC 5424 syslog feeds into your SIEM — Splunk, Wazuh, Graylog, ELK, Sentinel. Continuous visibility into every blocked and allowed tool call.

→ SIEM events
Enterprise Journey

This lifecycle maps directly to enterprise security workflows — from initial risk assessment through continuous enforcement to regulatory audit trails that satisfy SOC 2, HIPAA, and PCI-DSS auditors.

Initial Assessment Policy Deployment Continuous Enforcement Compliance Verification Audit Trail
Live Demo

See it in action.

AgentWard scans every tool your agent can reach, risk-rates them, detects dangerous skill chains (lateral data movement between tools), generates a policy, and wires enforcement — all in seconds.

agentward scan · agentward init
See It Running

The full lifecycle, step by step.

01 SCAN agentward scan
Risk findings
⚠ Skill chain detected
email-mgr → web-browser
lateral data movement risk
3 critical 4 high 1 medium 1 low
02 CONFIGURE agentward configure
agentward.yaml
6 rules generated 3 approval gates
03 ENFORCE agentward inspect
Audit log
3 blocked 1 approved 2 allowed
04 COMPLY agentward comply
HIPAA §164.312
2 required gaps auto-fix available
05 VERIFY agentward probe
Probe results
65 passed 3 warnings 0 failed
Compatibility

Works with the tools you already use.

Claude Desktop Claude Code Cursor Windsurf VS Code OpenClaw / ClawdBot OpenAI SDKSOON LangChainSOON CrewAISOON

Python 3.11+ · No API key required · Everything runs locally · Mac + Linux

Enterprise SIEM

Every audit event is written in RFC 5424 syslog format alongside the JSON Lines log — no configuration needed. Compatible with any SIEM or log shipper that reads standard syslog.

Splunk Wazuh Graylog ELK / Filebeat Microsoft Sentinel Fluentd rsyslog Any RFC 5424 shipper

Stop YOLOing your agent permissions.

Start verifying.

5 seconds to see what your AI agent's tools can actually do.

GitHub Repo
Apache 2.0 · Python 3.11+ · No API key · Runs locally · aditya@agentward.ai
agentward — full walkthrough