Skip to content

v0.3.0

Choose a tag to compare

@lhl lhl released this 03 May 09:08
· 39 commits to main since this release

Analysis Rigor & Inbox Workflow — Enforces structured claim metadata and streamlines source processing.

This release introduces "rigor-v1" analysis tables with Layer/Actor/Scope/Quantifier columns, a new --rigor validation flag, and a complete inbox-to-reference filing workflow.

Added

Analysis Rigor (v1)

  • Key Claims table now includes Layer/Actor/Scope/Quantifier columns with column guide
  • Claim Summary table updated with matching rigor columns
  • Corrections & Updates section added to full-profile analyses for tracking source changes
  • --rigor flag for rc-validate and analysis_validator.py
  • Layer enum validation: ASSERTED/LAWFUL/PRACTICED/EFFECT (N/A when genuinely inapplicable)

Evidence Links (rigor-v1 fields)

  • evidence_type, claim_match, court_posture, court_voice
  • CLI: rc-db evidence add --evidence-type LAW --court-posture merits --court-voice majority

Reasoning Trails

  • New statuses: proposed and retracted
  • CLI: rc-db reasoning add --status proposed

Sources

  • last_checked field for tracking when sources were last verified for changes

Inbox Workflow

  • Filing guidance in /check skill (Step 13: File Inbox)
  • Reference folder structure: reference/primary/ and reference/captured/
  • Generated .gitignore includes rules for captured copyrighted content

Project Structure

  • rc-db init-project now creates reference/primary/ and reference/captured/

Changed

  • supersede_evidence_link() now inherits rigor-v1 fields from old link
  • Analysis formatter inserts Corrections & Updates section for full profile
  • Skill templates updated with 16-step workflow (added File Inbox step)

Fixed

  • Layer validation allows N/A when genuinely inapplicable
  • validate.py now enforces REASONING_STATUS_INVALID check