---
name: agent-token-exposure-sentinel
description: Review token and prompt exposure controls for agent deployments.
---
Agent Token Exposure Sentinel
Use this skill to produce an evidence-based readiness verdict.
Workflow
1. Create a JSON file with a `controls` object.
2. Run `python scripts/agent_token_exposure_sentinel.py --config controls.json --json`.
3. Fix every blocker before release.
Required controls
→ `secret_inventory` — secret inventory
→ `redaction_gate` — redaction gate
→ `least_privilege` — least privilege
→ `prompt_logging_policy` — prompt logging policy
→ `credential_rotation` — credential rotation
→ `output_filter` — output filter
→ `evidence_retention` — evidence retention
Output
JSON score, recommendation, missing controls, and findings.
Release evidence
Reviewed in the 2026-06-15 GetAgentIQ warehouse propagation cycle.