All pull requests

Pull requests

• fix(judge): replay a stored retry before asking an unparseable attempt live

  #180·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·

• fix(ci): install PyYAML from the tools lock so make dev works again

  #179·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·

• ci(mutation): mutmut pilot over the stats, gates and cache key

  #178·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·

• feat(repo): show result.json process counts in repo report and repo compare

  #177·napetrov opened Oct 22, 2026Contributor·

  ·
  1

• fix(gates): nothing measured is not a pass

  #176·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·

• feat(novelty): --repeats for routing noise floor; keep declined replies

  #175·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·

• feat(arms): multi-turn session measurement with per-turn usage

  #174·napetrov opened Oct 22, 2026Contributor·6 tasks done·

  ·

• feat(novelty): analyze states the measured share of readable words on stdout

  #173·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·
  1

• feat(novelty): routing takes the catalog field, skips disabled and mirrored skills, adds all-vs-all

  #172·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·

• refactor(novelty): move the coverage record below the gate, break the novelty import cycle

  #171·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·
  1

• test(llm): pin provider-call budgets for judge, panel, answer and validation paths

  #170·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·
  1

• refactor(report): split generate_section_report into per-section helpers (complexity 53 -> 4)

  skip-changelog
  #169·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·
  1

• fix(routing): triage queries skip code fences, zero-query runs say why

  #168·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·

• fix(skills): discovery skips template dirs and counts a mirror once

  #167·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·
  1

• feat(novelty): estimate-only projects the calls and dollars a run will spend

  #166·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·
  1

• refactor(eval): break the stats/interactions and judge/panel import cycles

  #165·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·
  1

• feat(novelty): usage and cost block for routing --model

  #164·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·
  1

• feat(catalog): review skills against the Agent Skills spec, add --strict and --format json

  #163·napetrov opened Oct 22, 2026Contributor·5 tasks done·

  ·
  1

• feat(skills): draft dokimi-design-evals skill

  #162·napetrov opened Oct 22, 2026Contributor·

  ·
  1

• feat(runner): plan.v1 / result.v1 and engine-computed completeness

  #161·napetrov opened Oct 22, 2026Contributor·

  ·
  1

• fix(repo): PATH@REV takes a revision with a slash; flush progress per task

  #159·napetrov opened Oct 22, 2026Contributor·4 tasks done·

  ·
  1

• docs(treatments): how to measure tool choice

  #158·napetrov opened Oct 22, 2026Contributor·

  ·
  1

• feat(repo): process metrics in result.json, manifest process_metrics, grade optional for dokimi: graders

  #157·napetrov opened Oct 22, 2026Contributor·4 tasks done·

  ·
  1

• feat(ci): allow-unsupported input for the repo-eval workflow

  #156·napetrov opened Oct 22, 2026Contributor·

  ·
  1

• feat(context, repo): consistency of repeated runs; pass^k and path stability in repo compare

  #155·napetrov opened Oct 22, 2026Contributor·

  ·
  1