build-chronology

Builds a medical chronology from all analysed documents in a matter. Use when Mandy says 'build chronology', 'medical chronology', 'timeline', 'history of treatment', or needs a chronological view of

SKILL.md

GitHub

Build Chronology

Important

  • Manual-only. Resolve matter first.
  • Consult chronology-format.md for output structure.
  • Uses Tier 1 only — reads .extract.json files, not raw PDFs.

Instructions

Step 0: Resolve Matter

Follow _shared/resolve-matter.md.

Step 1: Check for Unanalysed Documents

List PDFs without .extract.json siblings. If found: "There are {n} unanalysed documents. Chronology will be incomplete. Proceed anyway, or /analyse-document first?"

Step 2: Collect Chronology Entries

Read ALL .extract.json files in matters/{id}/documents/ (Tier 1, ~500-800 tokens each). Collect chronology_entries arrays from each extract.

Step 3: Merge and Sort

Sort all entries by date. Flag duplicates (same date + type + source → keep richest).

Step 4: Gap Analysis

Identify: treatment gaps (3+ months for same body part), untreated conditions, CoC gaps, contradictions, missing impairment assessments, unexplained capacity changes, missing initial consultation.

Step 5: Generate

Follow chronology-format.md for output structure. Save to chronology/chronology-{date}.md.

Step 6: Update and Commit

Update matter.json. Update index. Log and commit.

Step 7: Present

Show entry count, date range, sources, gaps, contradictions, and recommended next steps. Proactive stage check + limitation check.

Error Handling

  • No extracts found: "No analysed documents. Run /analyse-document first."
  • No chronology entries in extracts: "Documents were analysed but no dateable events found. Check extracts manually."