iiRecord
Agentic AI Atlas · pi-mono coding-agent output guard
output-guard:pi-coding-agenta5c.ai
II.
OutputGuard JSON

output-guard:pi-coding-agent

Structured · live

pi-mono coding-agent output guard json

Inspect the normalized record payload exactly as the atlas UI reads it.

File · security/output-guards/pi-coding-agent-output-guard.yamlCluster · security
Record JSON
{
  "id": "output-guard:pi-coding-agent",
  "_kind": "OutputGuard",
  "_file": "security/output-guards/pi-coding-agent-output-guard.yaml",
  "_cluster": "security",
  "attributes": {
    "displayName": "pi-mono coding-agent output guard",
    "appliesTo": "all",
    "phase": "both",
    "detectorKinds": [
      "refusal-stop-reason",
      "secret-leak",
      "format-violation",
      "custom"
    ],
    "onTrigger": "emit-system-message",
    "severity": "warn",
    "description": "Streaming + post-call guard sitting in front of pi-mono's event-bus.\nInspects assistant_message_delta chunks and tool_result blocks before\nthey hit the user-facing channel, raising warnings or rewriting per\ndetector. Distinct from the compliance-side ContentPolicy — this is the\nruntime gate, not the policy declaration.\n"
  },
  "outgoingEdges": [],
  "incomingEdges": [
    {
      "from": "agent-runtime-impl:pi.runtime@current",
      "to": "output-guard:pi-coding-agent",
      "kind": "output_guard_applied_by",
      "attributes": {}
    }
  ]
}