Skip to content

chore(models): refresh capabilities from command-code 1.89.1 - #218

Merged
pierreraby merged 1 commit into
mainfrom
chore/catalog-1.89.1
Nov 15, 2026
Merged

pierreraby merged 1 commit into
mainfrom
chore/catalog-1.89.1

Conversation

@pierreraby

Copy link
Copy Markdown
Collaborator

Summary

Refresh the generated capability catalog from command-code@1.88.5 to 1.89.1, resolving the independent upstream drift reported by #217's catalog check.

  • Add image input, reasoning and published effort levels for claude-fable-5-2, google/gemini-4-carbon, meta/muse-ember and thinkingmachines/inkling-2-small:free.
  • Raise Inkling 2 Small's output limit to 131,072 tokens.
  • Remove retired reflection/beam:free and stealth/crouton-alpha metadata.
  • Add regression assertions for the additions/removals and adapt retired-model fixtures.
  • Keep pricing and endpoint routing unchanged.

Validation

  • Public pricing check: PASS for all 98 API models; no price drift.
  • Latest CLI catalog check: PASS against command-code@1.89.1.
  • Full npm test: PASS, with PI_LOCAL_REQUIRED=1 and OMP_COMPAT_REQUIRED=1 and COMMANDCODE_TRANSPORT unset; real Pi/OMP processes against mocks, three protocols on both hosts.
  • Entrypoint typecheck against installed Pi 1.5.2 declarations: PASS.
  • Formatting and git diff --check: PASS.
  • No paid API requests, production config changes or installs. Existing review-worktree dependencies were temporarily reused through a symlink, removed before commit.

LSP checks on changed TS files reported no errors (one confirmation inconclusive); the explicit compiler gates above are authoritative. Auxiliary workflow security findings concern unchanged baseline files, not this patch: no workflow or entrypoint source changes are included. This PR does not claim to resolve that separate workflow-hardening debt.

Current maintenance is being handled solo; this change is kept separate from #217's externally contributed OAuth fix. No release tag or publication is part of this PR.

@github-actions

Copy link
Copy Markdown
Contributor

Dependency Review

✅ No vulnerabilities or license issues or OpenSSF Scorecard issues found.

Scanned Files

None

@github-actions

Copy link
Copy Markdown
Contributor

Memory benchmark

Compared base 2a5c55d with PR head 87ea0eb on the same GitHub-hosted darwin runner. Lower values are better.

Metric Base PR PR − Base Change
Stable RSS 106.9 MiB ± 0.6 MiB 107.2 MiB ± 0.8 MiB +0.2 MiB ± 1.2 MiB +0.2%
Physical footprint 61.4 MiB ± 0.5 MiB 61.4 MiB ± 0.8 MiB +0.3 MiB ± 0.5 MiB +0.5%
Physical peak 73.5 MiB ± 0.4 MiB 73.9 MiB ± 2.2 MiB +0.6 MiB ± 1.0 MiB +0.8%

Interpretation: No metric shows a clear Base-to-PR difference beyond its measured run-to-run variation.

Extension overhead above pi baseline

Metric Base overhead PR overhead Difference
Stable RSS +5.6 MiB ± 1.5 MiB +6.1 MiB ± 0.9 MiB +0.2 MiB ± 1.2 MiB
Physical footprint +2.5 MiB ± 0.3 MiB +3.0 MiB ± 0.8 MiB +0.3 MiB ± 0.5 MiB
Physical peak +3.8 MiB ± 0.8 MiB +4.4 MiB ± 1.9 MiB +0.6 MiB ± 1.0 MiB

Values are medians of 6 alternating, paired runs. The value after ± is the median absolute deviation (MAD). Each process was sampled 12 times after a 2500 ms warm-up.

Environment: pi 0.84.4, Bun 1.4.4, darwin arm64. RSS comes from ps; physical footprint and peak come from macOS footprint.

This is a comparative signal, not a pass/fail threshold. GitHub-hosted runner noise can affect absolute values.

@pierreraby
pierreraby merged commit d3eda89 into main Nov 15, 2026
15 checks passed
@pierreraby
pierreraby deleted the chore/catalog-1.89.1 branch November 15, 2026 11:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant