Skip to content

v0.1.10 - Harness Doctor v2 coupling diagnostics

Choose a tag to compare

View all tags
@baskduf baskduf released this 08 Jun 07:27
· 26 commits to main since this release
v0.1.10
d5cf04f

Patch release for Harness Doctor v2. This release turns Doctor from a flat baseline scan into a six-element repository health and coupling diagnostic while preserving the boundary between harness readiness and agent effectiveness.

Added

  • Harness Doctor v2 scoring across Instructions, Constraints, Feedback, Memory, Evaluation, and Governance.
  • First-class coupling findings for orphan constraints, orphan feedback, unoperationalized memory, unevaluated memory, ungoverned change types, and promotion gaps.
  • Optional Doctor gates for minimum score and critical coupling findings, disabled by default.
  • Decision memory for the Doctor v2 model and task outcome evidence for the implementation and review loop.

Changed

  • Update /harness doctor, the scoring rubric, example reports, component map, theory docs, and roadmap guidance to describe the six-element diagnostic.
  • Keep Proven/effectiveness signals unmeasured in Doctor output unless durable outcome evidence supports a claim.
  • Tighten feedback-binding heuristics so generic documentation mentions do not count as execution wiring, and unbound check scripts do not inflate Feedback health.
  • Expand Doctor regression tests for coupling findings, optional gates, non-comparable outcome evidence, illustrative effectiveness reports, and feedback-binding edge cases.
Assets 2
Loading