Part of Agent OS - Kernel-level governance for AI agents
GitHub CLI extension for Agent OS - The Linux Kernel for AI Agents
gh extension install imran-siddique/gh-agent-osgh agent-os run "analyze this codebase for security issues"gh agent-os audit --policy strictgh agent-os initgh agent-os status- strict: Maximum safety, requires human approval
- standard: Balanced safety and autonomy (default)
- permissive: Minimal restrictions
After running gh agent-os init, customize .agent-os/policy.yaml:
version: 1
name: my-policy
governance:
max_tokens: 4096
max_tool_calls: 10
timeout_seconds: 300
safety:
confidence_threshold: 0.8
require_human_approval: false
blocked_patterns:
- "rm -rf"
- "DROP TABLE"
allowed_tools:
- read_file
- write_file