data2dsl

Agent Plan: Ticket 044 - Koru Remediation Intent Generator

Context

semcod/koru operates on a closed feedback loop: DETECT → PLAN → EXECUTE → VERIFY → HEAL When data2dsl detects a CONFLICT or missing fact during VERIFY, koru requires a deterministic, machine-actionable remediation-intent feed specifying:

• Exact subject and metric in discrepancy.
• Required delta / repair instruction (synchronize_metric, restore_missing_entries, resolve_conflict).
• Pre-repair cryptographically pinned SHA-256 evidence digests.

Strategy

1. Implement RemediationIntentFormatter and format_remediation_intent() in src/data2dsl_remediation.py.
2. Add CLI subcommand feed-koru in src/data2dsl_cli.py.
3. Add full unit tests in tests/test_remediation_feed.py.
4. Validate ruff, pytest, and governance gate.