agent-governance - Govern AI agent tools with safety and policy controls
Provides patterns for enforcing policies, detecting dangerous intent, restricting tool usage, and auditing AI agent actions.
Tags
Updated: 2026-10-04Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Define composable governance policies
- Classify dangerous prompt intent
- Enforce tool access controls
- Apply content and rate limits
- Wrap tools with governance checks
- Record tool execution audits
- Load policies from YAML
Inputs
- Governance policy configuration
- Prompt or tool argument content
- Agent tool functions
- Optional audit trail
Outputs
- Allow, deny, or review decisions
- Intent threat signals
- Tool execution results or errors
- Audit trail entries
Requirements
- Python runtime
- Async-compatible tool functions
- PyYAML for YAML policy loading