feat: supercharge muse code predict for agents
- Add -j alias for --json - Add exit_code + duration_ms to _PredictJson envelope - Add _ExplainJson TypedDict: structured signal breakdown for --explain --json Agents can now parse per-signal scores, reasons, and co-change partners for a single symbol without screen-scraping human output - Fill docstrings: _sanitise, _pct_bar, _confidence_label, _iter_symbol_ops, _build_predictions (already had one), _PredictJson, _ExplainJson, run() - Wire start_timer() throughout
TDD: 86 new tests in tests/test_predict_supercharge.py (zero prior coverage) Unit: _sanitise, _module_key, _pct_bar, _confidence_label, _iter_symbol_ops, _build_predictions (12 cases) Integration: json schema, -j alias, exit_code/duration_ms, filters, --explain human + json, empty repo Security: --min-confidence bounds, --explain format validation
0 comments
muse hub commit comment sha256:58f652da30b4f96e52b7caaceb00e6532e6edf3802e80975f6d9b1d9df9c2993 --body "your comment"
No comments yet. Be the first to start the discussion.