{
  "id": "scan:2026-10-23T04:49:56Z",
  "schema_version": 1,
  "status": "partial",
  "started_at": "2026-10-23T04:49:56Z",
  "finished_at": "2026-10-23T04:51:48Z",
  "run_url": "https://github.com/reacher-z/Awesome-CUA/actions/runs/39902403603",
  "schedule": {
    "timezone": "UTC",
    "times_local": [
      "00:17",
      "08:17",
      "16:17"
    ]
  },
  "before": {
    "resources": 3251,
    "bibliographic_records": 2767,
    "arxiv_archive_records": 2905,
    "excluded_records": 138,
    "curated_papers": 1200,
    "curated_resources": 471,
    "candidates": 4293
  },
  "after": {
    "resources": 3251,
    "bibliographic_records": 2767,
    "arxiv_archive_records": 2905,
    "excluded_records": 138,
    "curated_papers": 1200,
    "curated_resources": 471,
    "candidates": 4297
  },
  "new_papers": [],
  "new_candidate_count": 4,
  "pending_metadata_changes": 3,
  "source_fetches": {
    "successful": 54,
    "total": 54
  },
  "paper_searches": {
    "successful": 32,
    "total": 32,
    "unique_identifiers": 1496,
    "new_candidates": 0
  },
  "candidate_observations": {
    "started_at": "2026-10-23T04:50:43Z",
    "finished_at": "2026-10-23T04:51:19Z",
    "ok": true,
    "exit_code": 0,
    "error": null,
    "state": "rate_limited",
    "complete": false,
    "scan_status": "rate_limited",
    "run_id": "run:2eacb644a5a5cbc46426d4c9",
    "request_attempted": true,
    "successful_responses": 1,
    "requests_attempted": 3,
    "new_unique_repositories": 7,
    "total_unique_repositories": 9310,
    "pending_queries": 17,
    "unknown_requests": 2,
    "epoch_id": "epoch:618133feb6ed9ebb26b1d40d",
    "query_status_counts": {
      "observed_pages_exhausted": 84,
      "pending": 17,
      "split": 23,
      "superseded_input": 19
    },
    "rejected_rows": {},
    "warnings": [
      "GitHub candidate observations paused for provider rate limiting; verified progress and cursor are retained.",
      "2 GitHub candidate requests have no accepted metadata response."
    ],
    "files": {
      "github-candidates.jsonl": "103c282ddc210c699955dcacf67f49262c4a12a6f0622168cb448b933ae3f4d1",
      "github-candidates-manifest.json": "50a21bcaa1f6c2280b2d05957b498dd644c8e15bdb145f5f784f8342f028052b",
      "github-candidate-progress.json": "cb016339726e6ec3a424a851ad6762e5192c75c4fcb2f0e1efa0fb58295128cd"
    },
    "published": true,
    "network_reservations": 3,
    "staged_files": {
      "github-candidates.jsonl": "103c282ddc210c699955dcacf67f49262c4a12a6f0622168cb448b933ae3f4d1",
      "github-candidates-manifest.json": "50a21bcaa1f6c2280b2d05957b498dd644c8e15bdb145f5f784f8342f028052b",
      "github-candidate-progress.json": "cb016339726e6ec3a424a851ad6762e5192c75c4fcb2f0e1efa0fb58295128cd"
    },
    "comparison_context_rebound": true,
    "durable_validation": {
      "ok": true,
      "checked_at": "2026-10-23T04:51:48Z",
      "mode": "resumable_triplet",
      "required": true,
      "files": {
        "github-candidates.jsonl": "103c282ddc210c699955dcacf67f49262c4a12a6f0622168cb448b933ae3f4d1",
        "github-candidates-manifest.json": "50a21bcaa1f6c2280b2d05957b498dd644c8e15bdb145f5f784f8342f028052b",
        "github-candidate-progress.json": "cb016339726e6ec3a424a851ad6762e5192c75c4fcb2f0e1efa0fb58295128cd"
      },
      "repositories": 9310
    }
  },
  "phases": [
    {
      "name": "discovery",
      "started_at": "2026-10-23T04:49:56Z",
      "finished_at": "2026-10-23T04:50:07Z",
      "ok": true,
      "exit_code": 0,
      "error": null
    },
    {
      "name": "huggingface",
      "started_at": "2026-10-23T04:50:07Z",
      "finished_at": "2026-10-23T04:50:40Z",
      "ok": true,
      "exit_code": 0,
      "error": null
    },
    {
      "name": "recent_arxiv",
      "started_at": "2026-10-23T04:50:40Z",
      "finished_at": "2026-10-23T04:50:40Z",
      "ok": false,
      "exit_code": 0,
      "error": null,
      "complete": false,
      "new_count": 0,
      "warnings": [
        "HTTP Error 429: Unknown Error"
      ],
      "cache_used": false,
      "rate_limited": true
    },
    {
      "name": "harvest",
      "ok": false,
      "deferred": true,
      "error": "Deferred after the recent arXiv query returned HTTP 429; discovered identifiers remain queued."
    },
    {
      "name": "github_candidates",
      "started_at": "2026-10-23T04:50:43Z",
      "finished_at": "2026-10-23T04:51:19Z",
      "ok": true,
      "exit_code": 0,
      "error": null,
      "state": "rate_limited",
      "complete": false,
      "scan_status": "rate_limited",
      "run_id": "run:2eacb644a5a5cbc46426d4c9",
      "request_attempted": true,
      "successful_responses": 1,
      "requests_attempted": 3,
      "new_unique_repositories": 7,
      "total_unique_repositories": 9310,
      "pending_queries": 17,
      "unknown_requests": 2,
      "epoch_id": "epoch:618133feb6ed9ebb26b1d40d",
      "query_status_counts": {
        "observed_pages_exhausted": 84,
        "pending": 17,
        "split": 23,
        "superseded_input": 19
      },
      "rejected_rows": {},
      "warnings": [
        "GitHub candidate observations paused for provider rate limiting; verified progress and cursor are retained.",
        "2 GitHub candidate requests have no accepted metadata response."
      ],
      "files": {
        "github-candidates.jsonl": "103c282ddc210c699955dcacf67f49262c4a12a6f0622168cb448b933ae3f4d1",
        "github-candidates-manifest.json": "50a21bcaa1f6c2280b2d05957b498dd644c8e15bdb145f5f784f8342f028052b",
        "github-candidate-progress.json": "cb016339726e6ec3a424a851ad6762e5192c75c4fcb2f0e1efa0fb58295128cd"
      },
      "published": true,
      "baseline_files": {
        "github-candidates.jsonl": "cecbfcedd8ef144b8052a2564a55cfa56f41e599eda8ed0b427f3705efa90a8f",
        "github-candidates-manifest.json": "0e6dc266e95580c0bf1ec428dcbdb7e5b5ed1d9ad6a3d65fd5616af76065f903",
        "github-candidate-progress.json": "595997da8ad3a4b0f253debe96874c0e90ad33b144c321f54a98627557a788c7"
      },
      "ready_to_publish": false,
      "network_reservations": 3,
      "staged_files": {
        "github-candidates.jsonl": "103c282ddc210c699955dcacf67f49262c4a12a6f0622168cb448b933ae3f4d1",
        "github-candidates-manifest.json": "50a21bcaa1f6c2280b2d05957b498dd644c8e15bdb145f5f784f8342f028052b",
        "github-candidate-progress.json": "cb016339726e6ec3a424a851ad6762e5192c75c4fcb2f0e1efa0fb58295128cd"
      },
      "comparison_context_rebound": true,
      "durable_validation": {
        "ok": true,
        "checked_at": "2026-10-23T04:51:48Z",
        "mode": "resumable_triplet",
        "required": true,
        "files": {
          "github-candidates.jsonl": "103c282ddc210c699955dcacf67f49262c4a12a6f0622168cb448b933ae3f4d1",
          "github-candidates-manifest.json": "50a21bcaa1f6c2280b2d05957b498dd644c8e15bdb145f5f784f8342f028052b",
          "github-candidate-progress.json": "cb016339726e6ec3a424a851ad6762e5192c75c4fcb2f0e1efa0fb58295128cd"
        },
        "repositories": 9310
      }
    },
    {
      "name": "build",
      "started_at": "2026-10-23T04:51:22Z",
      "finished_at": "2026-10-23T04:51:23Z",
      "ok": true,
      "exit_code": 0,
      "error": null
    },
    {
      "name": "validate",
      "started_at": "2026-10-23T04:51:39Z",
      "finished_at": "2026-10-23T04:51:45Z",
      "ok": true,
      "exit_code": 0,
      "error": null
    }
  ],
  "warnings": [
    "recent_arxiv: failed",
    "HTTP Error 429: Unknown Error",
    "harvest: Deferred after the recent arXiv query returned HTTP 429; discovered identifiers remain queued.",
    "GitHub candidate observations paused for provider rate limiting; verified progress and cursor are retained.",
    "2 GitHub candidate requests have no accepted metadata response."
  ],
  "policy": "Automated scans discover candidates and verify bibliographic metadata. Editorial approval, topic annotations, and license claims require separate source review.",
  "network_budget": {
    "round_id": "187c92f20109458c921396b9c2ee99b9",
    "max_requests": 180,
    "max_seconds": 600,
    "requests_reserved": 91,
    "responses_received": 91,
    "unknown_attempts": 0,
    "by_phase": {
      "discovery": 55,
      "huggingface": 32,
      "recent_arxiv": 1,
      "github_candidates": 3
    },
    "by_provider": {
      "raw.githubusercontent.com": 52,
      "iaie-physicalai.github.io": 2,
      "huggingface.co": 33,
      "export.arxiv.org": 1,
      "api.github.com": 3
    },
    "provider_cooldowns": {
      "api.github.com": "2026-10-23T04:52:05.438587+00:00",
      "export.arxiv.org": "2026-10-23T04:51:40.724747+00:00"
    },
    "blocked_providers": [
      "api.github.com",
      "export.arxiv.org"
    ],
    "ledger_sha256": "d030be1f45852863c938dd8b16fb29259483004040ba95d1d4d7246479cafa14",
    "count_boundary": "Reservations are persisted before I/O; interrupted attempts remain conservatively charged. HTTP status observation does not establish complete or validated content."
  }
}