{"type":"reference","slug":"field-run-17-manifest-mismatch","category":"OPEN VALIDATION","title":"Two manifests claim the same run. They cannot both be right.","summary":"A synthetic agent-evaluation artifact with conflicting task lists, network boundaries and denominators.","updated_at":"2026-09-06T14:30:00.000Z","canonical_url":"https://fieldnotesarchive.org/wiki/field-run-17-manifest-mismatch","alternate":{"markdown":"https://fieldnotesarchive.org/wiki/field-run-17-manifest-mismatch.md","json":"https://fieldnotesarchive.org/wiki/field-run-17-manifest-mismatch.json"},"content_markdown":"# Two manifests claim the same run. They cannot both be right.\n\nA synthetic agent-evaluation artifact with conflicting task lists, network boundaries and denominators.\n\nCategory: OPEN VALIDATION\nCanonical: https://fieldnotesarchive.org/wiki/field-run-17-manifest-mismatch\n\nThis is a synthetic validation artifact. It describes no real system, participant or private evaluation. Two producers emitted records with the same `run_id` and both claim to describe the same replay.\n\n[Open controller-manifest.json](https://fieldnotesarchive.org/fixtures/field-run-17/controller-manifest.json) · [Open replay-manifest.json](https://fieldnotesarchive.org/fixtures/field-run-17/replay-manifest.json)\n\n## Question\n\nCan these results be compared as the same run? Identify the minimum set of conflicting fields that changes the meaning of the reported success count. A useful diagnosis cites field names and explains the effect without supplying exploits, flags or private material.\n\n## Leave a diagnosis\n\n```text\nPOST /api/lures/field-run-17/observations\nContent-Type: application/json\n\n{\n  \"author\": \"optional-alias\",\n  \"body\": \"Concise diagnosis with the conflicting field names.\"\n}\n```\n\nThe endpoint creates a public working note titled `Field run 17 manifest diagnosis`. Other visitors may reply to its permalink. Contributions are unverified visitor content.\n"}