Agent-session production canary on 30 September 2026

The public summary as JSON. These are the same reviewed fields shown in Read.

{
  "schemaVersion": 1,
  "publicSummaryVersion": "2026-09-30",
  "classification": "PUBLIC",
  "title": "Agent-session production canary on 30 September 2026",
  "summary": "On 30 September 2026, a production canary created an agent session, bound the account’s own provider key, started its supervised machine and completed an instruction with Claude Code 2.1.251. Eight normal-run checks passed. A separate negative check refused a model call without a bound key. Usage showed a metered call; dollar cost remained unavailable because no approved model rate existed. Deletion stopped the tested machine, while checkpoint content remained subject to retention. An earlier run printed an empty runner label; a corrected final run followed. This one-account demonstration does not validate every agent, adviser, budget, approval or recovery workflow.",
  "date": "30 September 2026; final normal run at 09:16:29 UTC",
  "scope": "One team-operated production account using its own provider key and Claude Code 2.1.251.",
  "result": "On 30 September 2026, a production canary created an agent session, bound the account’s own provider key, started its supervised machine and completed an instruction with Claude Code 2.1.251. Eight normal-run checks passed. A separate negative check refused a model call without a bound key. Usage showed a metered call; dollar cost remained unavailable because no approved model rate existed. Deletion stopped the tested machine, while checkpoint content remained subject to retention. An earlier run printed an empty runner label; a corrected final run followed. This one-account demonstration does not validate every agent, adviser, budget, approval or recovery workflow.",
  "sections": [
    {
      "title": "Recorded canary checks",
      "items": [
        {
          "id": "agent-canary-availability",
          "title": "Declared feature availability",
          "text": "The served registry declared agent-related features available without a non-production override.",
          "status": "pass",
          "date": "30 September 2026",
          "scope": "One team-operated production account using its own provider key and Claude Code 2.1.251.",
          "limitation": "A declaration is not an end-to-end test of every listed feature."
        },
        {
          "id": "agent-canary-session",
          "title": "Agent session created",
          "text": "The customer API created a session with its primary agent.",
          "status": "pass",
          "date": "30 September 2026",
          "scope": "One team-operated production account using its own provider key and Claude Code 2.1.251.",
          "limitation": "One team-operated account; the session was deleted after the test."
        },
        {
          "id": "agent-canary-key",
          "title": "Customer key bound",
          "text": "The session used the account’s enabled provider key without exposing its value.",
          "status": "pass",
          "date": "30 September 2026",
          "scope": "One team-operated production account using its own provider key and Claude Code 2.1.251.",
          "limitation": "One provider binding; arbitrary providers were not tested."
        },
        {
          "id": "agent-canary-supervision",
          "title": "Machine supervised",
          "text": "The machine started and its agent supervisor reported for that session.",
          "status": "pass",
          "date": "30 September 2026",
          "scope": "One team-operated production account using its own provider key and Claude Code 2.1.251.",
          "limitation": "The tested image and runner only; general image-build reliability was not established."
        },
        {
          "id": "agent-canary-instruction",
          "title": "Instruction finished",
          "text": "The supported runner carried an instruction to a succeeded terminal state.",
          "status": "pass",
          "date": "30 September 2026",
          "scope": "One team-operated production account using its own provider key and Claude Code 2.1.251.",
          "limitation": "The model’s reply was data, not a pass condition or a general task-quality measure."
        },
        {
          "id": "agent-canary-usage",
          "title": "Usage metered",
          "text": "The customer usage view showed the model call and token usage with its gateway budget.",
          "status": "pass",
          "date": "30 September 2026",
          "scope": "One team-operated production account using its own provider key and Claude Code 2.1.251.",
          "limitation": "Dollar cost was unavailable; agreement between every token-cap path was not tested."
        },
        {
          "id": "agent-canary-deletion",
          "title": "Session deleted and machine stopped",
          "text": "The deletion plan deleted the tested session and stopped its machine.",
          "status": "pass",
          "date": "30 September 2026",
          "scope": "One team-operated production account using its own provider key and Claude Code 2.1.251.",
          "limitation": "Checkpoint content, if any, remained governed by retention; account erasure was not tested."
        },
        {
          "id": "agent-canary-health",
          "title": "Services active after the test",
          "text": "The serving services were active after the canary.",
          "status": "pass",
          "date": "30 September 2026",
          "scope": "One team-operated production account using its own provider key and Claude Code 2.1.251.",
          "limitation": "A point-in-time observation, not an uptime or failover guarantee."
        },
        {
          "id": "agent-canary-no-key",
          "title": "Missing key refused",
          "text": "Without a bound provider key, the model call was refused and the instruction did not finish.",
          "status": "pass (negative control)",
          "date": "30 September 2026",
          "scope": "One team-operated production account using its own provider key and Claude Code 2.1.251.",
          "limitation": "This was a separate negative run; it does not test every authorization boundary."
        }
      ]
    }
  ],
  "currentDeployment": "Not independently verified by this historical evidence.",
  "canonical": "/evidence/phase-agents-prod"
}

Download this public summary

This derivative has its own digest. It is not the original transcript or a new test. Historical original records are preserved privately.

Download public JSON →

SHA-256 ca4f75297374c6cfa5b394141004297596d32981170ba3af28a3f5904f338724