2026-05-18: Rule-of-5-universal review on full
implementation; found 11 issues (5 P0, 4 P1, 2 P2); ran issue-review to
sharpen acceptance criteria; fixed all 5 P0 correctness bugs:
exported-threshold violations (pretender-2rd), ABC threshold enforcement
(pretender-xgz), file_lines_max check (pretender-4ow), cognitive
complexity sequence_id dedup (pretender-9r1), lines_code/lines_comment
classification (pretender-3s5)
2026-05-19: Implemented pretender-oyg
— min_assertions metric for test role (TDD pipeline: assertion counting
per CodeUnit, Python assert/self.assert*/pytest.raises/pytest.warns
captures); implemented pretender-4eh — ABC smell weights
engine (call_weights threaded through QueryEngine, Python plugin.toml
gets eval=5/exec=5/compile=3, JS/TS parsers updated); both closed and
pushed
Arc: Went from post-P0-fix state to two new feature
tickets complete; engine now scores weighted calls and flags
under-asserted tests
Next: Remaining P1s — pretender-5wh (merge
get_parser/registry refactor), pretender-lo5 (exclude patterns),
pretender-0cc (validate_percent false-positive triage); then P2s
pretender-tao and pretender-dx1
Session 16:51 (miblioteca)
Rule of 5 Universal review of CaptureView/tracer architecture
diagnostic (5 stages, 14 issues found across correctness, clarity,
feasibility, completeness)
Closed mibilioteca-rukz: extracted
TuningConfig to src/shared/tuningConfig.ts,
breaking the sensors ↔︎ ghost import cycle (commit
519ddbc)
Closed mibilioteca-n1cw: added
src/tracer/index.ts barrel exports, migrated all external
tracer imports to public API (commit 5a1384b)
Closed mibilioteca-47fx: extracted
IMUSteadinessMonitor from CaptureView via TDD
— 19 new tests, 876 total pass; CaptureView dropped 6
sensor imports, 4 private methods, 4 private fields (commit
e4aab9d)
Decision: preferred Option A (relocate type file) over Option B
(constructor injection) for TuningConfig — zero API changes
vs. callsite propagation
Next: run bd ready for next ticket;
remaining open specs include add-ghost-calibration (1 task) and
add-ghost-tuning-panel (2 tasks)
Sessions 16:51 (dont — 3
sessions)
2026-05-13: Implemented managed docs support
(dont/AGENTS.md, doctor checks,
--fix repair); closed epic dont-ymdq; pushed
commits
2026-05-18: Read LLM evaluation report; created
OpenSpec add-eval-readiness with 3 new specs
(dont-analytics, dont-ephemeral-mode,
dont-eval-export); ran Ro5 review and applied all
fixes
Arc: Managed-docs epic → eval-readiness specs →
large bug-fix pass → import/test hardening; import system now has
canonical IDs and verified adapter contracts
Next: Implement eval-readiness specs (start with
dont-analytics); or continue CLI-CONSISTENCY sweep
(dont-4ian, dont-6os6,
dont-1afj)