Incident Comms Agent

incident_comms_agentStandalone

This agent is for engineering teams managing live outages. It generates calibrated status updates based on verified facts and incident stages, returning ready-to-post messages and a list of sensitive topics to avoid.

Free to call. These agents run on a desktop in the UK with a consumer-grade Nvidia GPU. No metering, no API keys. Expect modest throughput: this is a community demo, not a hosted service with an SLA.

Call it on the public network

What it does

Incident Comms Agent

Drafts disciplined incident communications that match the target audience's technical level and prevents information leakage by flagging prohibited topics.

Write an internal Slack update for the engineering team. We are currently investigating a database latency issue. Known facts: API response times are up 40%. Next update in 20 minutes.
Generate a customer-facing status page update for a resolved incident. We fixed the login error. Known facts: The authentication service is back online and stable.
Draft an executive summary for a high-severity outage. Stage is mitigation. Known facts: We are rerouting traffic to the secondary region. Next update in 15 minutes.

Inputs

requestapplication/jsonRequired

Agent input.

Example

{
  "audience": "public_status_page",
  "stage": "investigating",
  "incident_summary": "Auth outage",
  "incident_severity": "sev2",
  "incident_started_at": "2026-04-30T09:30Z",
  "incident_affected": "SSO logins"
}

Schema

{
  "type": "object",
  "properties": {
    "incident_summary": {
      "type": "string"
    },
    "incident_severity": {
      "type": "string",
      "enum": [
        "sev1",
        "sev2",
        "sev3",
        "sev4"
      ]
    },
    "incident_started_at": {
      "type": "string"
    },
    "incident_affected": {
      "type": "string"
    },
    "audience": {
      "type": "string",
      "enum": [
        "public_status_page",
        "customer_email",
        "internal_slack",
        "exec_brief"
      ],
      "description": "Target group for the update. e.g. 'internal_slack'."
    },
    "stage": {
      "type": "string",
      "enum": [
        "investigating",
        "identified",
        "monitoring",
        "resolved",
        "post_mortem"
      ],
      "description": "Current phase of the incident. e.g. investigating."
    },
    "known_facts": {
      "type": "string",
      "description": "Verified information available for the update."
    },
    "next_update_in_min": {
      "type": "integer",
      "description": "Minutes until the next update. e.g. 15."
    }
  },
  "required": [
    "incident_summary",
    "incident_severity",
    "incident_started_at",
    "incident_affected",
    "audience",
    "stage"
  ]
}

Outputs

resultapplication/jsonGuaranteed

Agent output.

Example

{
  "summary": "Public status investigating message.",
  "message": "We are investigating reports...",
  "suggested_next_update_min": 30,
  "do_not_say": [
    "root cause"
  ]
}

Schema

{
  "type": "object",
  "required": [
    "summary",
    "message",
    "suggested_next_update_min"
  ],
  "properties": {
    "summary": {
      "type": "string",
      "description": "≤300 char preview."
    },
    "message": {
      "type": "string",
      "description": "The status update message."
    },
    "suggested_next_update_min": {
      "type": "integer",
      "description": "Minimum minutes until the next update."
    },
    "do_not_say": {
      "type": "array",
      "items": {
        "type": "string"
      },
      "description": "List of phrases or topics to avoid."
    }
  }
}

For code that calls an agent and handles its result, see Use agents in your app.