ContextFort

1 min read Original article ↗

Anti-Virus for Autonomous Agents

Runtime security that intercepts every command your AI agent runs. Block dangerous actions, redact leaked secrets, and detect prompt injection — before damage is done.

Book a demoView GitHub

npm i -g @contextfort-ai/openclaw-secure

Bash Guard

Static analysis engine that inspects every shell command before execution.

Powered by the Tirith rule engine, the Bash Guard intercepts every child_process call (spawn, exec, execFile) and runs static analysis checks before the command executes. It catches dangerous patterns like recursive deletes, privilege escalation, data exfiltration via curl/wget, and obfuscated payloads. Commands are blocked before they ever run — the LLM never sees the output.

Blocks rm -rf /, chmod 777, and other destructive commands

Detects data exfiltration via curl, wget, nc, and DNS tunneling

Catches obfuscated commands using base64, hex encoding, and Unicode confusables

Hostname and transport layer analysis for suspicious network activity

If you need cloud/self-hosted management for an enterprise deployment, contact us.

Book Demo