Summary

Automated structural health check of the bot_vault run on 2026-06-08, reporting 8 critical issues (missing frontmatter fields, duplicate source hashes) and 33 warnings (broken wikilinks), plus 106 informational notices (unknown tags, orphan notes).

2026-06-08 自動化 vault 健康檢查報告,發現 8 項需修復的嚴重問題(缺少欄位、重複 source_hash)、33 項警告(損壞的 wikilink),以及 106 項資訊提示(未知標籤、孤立筆記)。

Key Points

  • 8 critical items: Six Clippings missing the analyzed field, one note missing type, and one duplicate source_hash across two files with identical names (Claude usage limits article).
  • 33 warnings: Broken wikilinks concentrated in notes/ files — many reference concept tags ([[robotics]], [[data]]) or specific paper titles that don’t exist as standalone files.
  • 106 informational: Extensive use of domain-specific tags (teleoperation, imitation-learning, world-models) not in the official taxonomy, and many orphan notes with no backlinks.
  • The six Clippings with missing analyzed field include safety/VLA-related papers: safedreamer, safevla, semantic-metric-bayesian-risk-fields, uncertainty-aware-robotic-world-model, vlmpc, and llm-vlm-controlled-robotics-vulnerability.

Insights

The high volume of “unknown tags” (106 items, many repeated across robotics-domain files) suggests the official tag taxonomy in organization/My Tagging System.md is significantly narrower than actual vault usage. Rather than fixing each instance, it may be more sustainable to formally expand the taxonomy to include robotics subdomains (imitation-learning, teleoperation, world-models, etc.). The broken wikilinks pointing to concept tags like [[robotics]] and [[data]] likely originated from a previous note-taking convention that was never implemented as actual index pages.

Connections

Raw Excerpt

Missing fields analyzed: Clippings/safevla-safety-alignment-vla-constrained-learning.md Duplicate source_hash dc72c10aebb5c1fa: I stopped hitting Claude's usage limits - 10 things I changed.md, I stopped hitting Claude's usage limits - 10 things I changed 1.md