Skip to content

Phase 3: 19 AC-aligned rules + generic text_features extractor #50

@MTCMarkFranco

Description

@MTCMarkFranco

Goal: Close the §5 backlog from \docs/comparison/lambda-rag-vs-air-canada.md\ — author 19 priority AC rules and add a domain-agnostic numeric-feature extractor so any future ruleset can reference numeric thresholds without hardcoding.

Scope:

  • Author 19 new rules in \samples/contracts/ac-demo-ruleset.json\ (payment, IP, liability carve-outs, insurance, crypto, privacy, AI, subcontracting, service-location, Quebec).
  • Add \TextFeatureExtractor\ (regex over English prose) emitting \ ext_features.{day_counts,month_counts,year_counts,percent_values,dollar_amounts}\ + _min/_max\ on every section.
  • Bump topic-map \contract.v1.json\ to v1.1.0 (tax, subcontracting, ai, service_locations).
  • Engine must remain domain-agnostic — proven by tests using synthetic non-AC rulesets.

Acceptance: All existing corpus goldens unchanged in outcome; AC end-to-end produces real findings (no hallucinations); ≥10 new tests including non-AC genericness coverage.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementNew feature or requestgovernanceRule governance, audit, idempotencyphase-1-pattern-defPhase 1: Pattern definition (writing)

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions