Citation Fixer

Provenance audit for knowledge base notes. Scans notes for citation compliance and fixes formatting, missing dates, missing sources, and broken reference links.

Contract

• Every note is scanned for citation compliance
• Missing citations flagged with specific location and suggested fix
• Malformed citations fixed to the standard format
• Results reported with counts (scanned, fixed, remaining gaps)
• Read-only by default: draft fix suggestions unless the user approves batch apply
• Never fabricate a citation for an uncited claim — flag it

Citation Standard

Every fact in the knowledge base should carry an inline [Source: ...] citation:

• User's own statements: [Source: user, {context}, YYYY-MM-DD]
• Web content: [Source: {publication}, {URL}, YYYY-MM-DD]
• Papers/articles: [Source: {authors}, "{title}", {year}]
• Conversations: [Source: chat with {name} re: {topic}, YYYY-MM-DD]
• Synthesis: [Source: compiled from {list of source slugs}]
• AI model output: [Source: {model name}, YYYY-MM-DD]

Phases

1. Scan. Walk through source notes, concept notes, and wiki pages. Check for inline [Source: ...] citations on factual claims.
2. Identify issues:
   • Facts without any citation
   • Citations missing date
   • Citations missing source type
   • Citations with wrong format
   • Reference links that 404 or point to wrong slugs
3. Draft fixes. For each issue, show the current text and the proposed fix. Group by note.
4. Report. Count: notes scanned, citations found, issues fixed, remaining gaps. Surface as a queue item for user review.

Output Format

Citation Audit — YYYY-MM-DD
────────────────────────────
Notes scanned:       X
Citations found:     X total (X valid, X issues)
Issues by type:
  Missing citation:     X
  Missing date:         X
  Missing source type:  X
  Malformed format:     X
  Broken reference:     X
────────────────────────────
Top notes to fix: [paths]

Anti-Patterns

• Fabricating a citation for an uncited claim — flag it as missing, don't invent
• Overwriting a claim citation without checking the source still exists
• Being too aggressive with ad-hoc conversational facts that don't need formal provenance (preference, opinion, speculation clearly marked as such)
• Running the full audit on the entire knowledge base without prioritizing — start with high-traffic pages
• Editing notes directly without the user's approval on the proposed changes