aport-agent-guardrail - Pre-action authorization for AI agents
Evaluates AI agent tool calls against passports and policies before execution, blocking unauthorized actions and recording decisions.
Tags
Updated: 2026-09-28Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Evaluate tool calls
- Enforce passport policies
- Block unauthorized actions
- Check allowlists
- Apply rate limits
- Log audit decisions
Inputs
- Tool call context
- Agent passport
- Policy configuration
- APort API settings
Outputs
- ALLOW or DENY decisions
- Permitted tool execution
- Blocked tool execution
- Decision logs
- Audit trails
- Signed API receipts
Requirements
- Node.js 18 or later
- jq
- Compatible runtime with before_tool_call hook support