SE
Skill Eval

Changelog

1.5.0 (2026-07-02)

Features

  • import-agit: add –origin filter, –dry-run preview, quality metadata in output (#31) (1bbc7cf)

1.4.0 (2026-07-01)

Features

  • context propagation — ctx through CmdBuilder, per-invocation timeout, graceful cancellation (#14) (756e8b8)
  • import-agit: integrate with agengit steps/eval, –merge, source metadata, smart filtering (#26) (164295f)

Bug Fixes

  • clear golangci-lint findings, split import_agit.go into internal/agit, fix CI (#28) (9d9e85b)
  • collect errors in cmdRun instead of aborting; pass –model/–skill to generic agent (#22) (dc579a2)
  • make CmdBuilder a parameter instead of package-level var to prevent races (#23) (f6e5756), closes #17
  • remediate all 22 security vulnerabilities from audit (4aefb54)
  • remove copilot from schema enum, use @v2 refs in release.yml (#21) (df7c6e0)

1.3.0 (2026-06-30)

Features

  • context propagation — ctx through CmdBuilder, per-invocation timeout, graceful cancellation (#14) (756e8b8)
  • import-agit: integrate with agengit steps/eval, –merge, source metadata, smart filtering (#26) (164295f)

Bug Fixes

  • clear golangci-lint findings, split import_agit.go into internal/agit, fix CI (#28) (9d9e85b)
  • collect errors in cmdRun instead of aborting; pass –model/–skill to generic agent (#22) (dc579a2)
  • make CmdBuilder a parameter instead of package-level var to prevent races (#23) (f6e5756), closes #17
  • remediate all 22 security vulnerabilities from audit (4aefb54)
  • remove copilot from schema enum, use @v2 refs in release.yml (#21) (df7c6e0)

1.2.0 (2026-06-27)

Features

  • import-agit converts recorded sessions into evals.json corpora (6670222)

Bug Fixes

  • benchmark: correct best/worst ranking, populate run_summary, extract pi token counts (83a9354)
  • lint: remove unused cliJSONEnvelopeKey const flagged by golangci-lint (2fb9c82)

1.1.0 (2026-06-27)

Features

  • add –baseline-only flag to run and loop commands (94771cb)
  • add –llm-suggestions and high-baseline eval-quality hint (26cca28)
  • add HTML report command with actionable suggestions (3ca892c)

Bug Fixes

  • release: remove release-as pin so versions advance past 1.0.0 (f769cf1)
  • restore cwd in tests with t.Chdir to stop getwd failures (0a17eec)
  • use config/evals topology for benchmark aggregation (e93b122)

1.0.0 (2026-06-25)

Features

  • add –verbose flag and structured logging (e6e41b9)
  • add cross-iteration delta to benchmark.json (e85d0d7)
  • add deterministic file/text assertion matchers (5011275)
  • add iteration lockfile and –resume (b39f004)
  • validate config against JSON schema (cf9b3f4)

0.6.0 (2026-06-25)

Features

  • –models flag for cross-model benchmarking (9fd5b45)
  • pre-push-guard pi extension (500b005)
  • pre-push-guard shows live status widget while checks run (ff5e658)

Bug Fixes

  • use <video> not for auto-fixing demo mp4 (6d9c5e3)
  • use <video> not for cross-model demo mp4 (8cdceb8)
  • use agent:model pairs in all cross-model examples, regenerate VHS (fd1a283)
  • use proper <video></video> closing tags, agent:model pairs throughout cross-model guide (8d3cbf5)

0.5.0 (2026-06-25)

Features

  • –fix flag for auto-refinement loop (evaluator-optimizer pattern) (fa85108)

Bug Fixes

  • a11y: modern-web-guidance improvements (3111adf)
  • ignore err from CombinedOutput in fixEval (grading handles failures) (e4fff21)

0.4.0 (2026-06-25)

Features

  • add view transitions for smooth page navigation (bcfc382)
  • docs: reorder docs menu, rename README to Home, add GitHub link (7d92c20)
  • docs: wire full SEO stack (7f82d89)

Bug Fixes

  • docs: compile images at build time to avoid 404 _image endpoint (b8da6b2)
  • docs: put Home at the top of the docs menu (8d42852)
  • docs: remove stale wrangler deploy config and add pages deploy script (ea75a79)
  • menu layout, guide ordering, and larger GIFs (c11bb0e)
  • VHS tapes use harness scripts, type real commands not comments (7690633)

0.3.0 (2026-06-25)

Features

  • docs: add CSP header and optimize font loading (b02a168)
  • docs: add mobile hamburger navigation (4305673)
  • migrate documentation to Astro site and modular markdown (614e051)

Bug Fixes

  • docs: add custom 404 error page (a6f2118)

0.2.0 (2026-06-24)

Features

Bug Fixes

  • resolve golangci-lint errors (ac9abc3)