{
  "id": 1830851,
  "title": "The ledger said 763 assets. The platforms said about 400.",
  "url": "https://urgent.news/2026/08/18/the-ledger-said-763-assets-the-platforms-said-about-400",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-08-18T23:02:02.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/untactit/the-ledger-said-763-assets-the-platforms-said-about-400-a03"
  },
  "original_language": "en",
  "account": "A recent analysis revealed that a sync ledger, which tracks assets across multiple AI platforms, contained a significant discrepancy. According to the ledger, there were 763 assets, but upon manual verification, only around 400 of those were confirmed as genuine. This discrepancy was traced back to three independent failure modes: ghosts, misses, and false positives.\n\nFirstly, assets that were deleted on the platform but still existed within the database accounted for at least 173 rows. Secondly, assets that were present on the platform but missed by the database totaled at least 63 rows. Lastly, false positives, where platform-provided defaults were mistakenly counted as user assets, amounted to at least 109 rows.\n\nOne specific example highlighted the issue: a platform's skill list showed 100 entries in the database, but the user only had 4 installed skills. The remaining 96 were simply the platform's public catalog, shown to every user. Similarly, a memory collector stored two placeholder lines as memories, even though the user had 8 real memories saved. And 42 instruction rows pointed to non-existent local rule directories, adding to the confusion.\n\nThe audit revealed that these issues stemmed from various sources: platform-catalog pages, placeholder strings, stale caches, and even non-existent file paths. The sync process ran successfully, writing these entries into the ledger, but this does not guarantee their accuracy. The key takeaway is that a successful write does not ensure a truthful read.\n\nTo rectify this, the suggested solution is to implement a read-back audit as a standard part of the sync process. After any sync, the system should re-read the platform's data and perform a direct comparison of IDs in both directions. This approach would help identify discrepancies, including platform-provided defaults, which need to be treated as a separate category. Running this verification regularly, rather than just once, would help maintain the integrity of the ledger.",
  "summary": "A sync ledger fails silently. Ours said 763 assets. When we stopped trusting it and counted by hand — opening 21 AI platforms in browser tabs, calling each platform's own API from the page context, comparing ID by ID — roughly 400 of those were real. This is the audit log of that gap. The numbers are from one working account, measured on 2026-08-18. Nothing here is hypothetical. The setup The…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}