Repository navigation
chore(models): refresh stale model defaults (FeatureClipStudio) - #26
Merged
Merged
Conversation
…EMINI_JUDGE_MODEL lib/judge-model.mjs exports JUDGE_MODEL (GEMINI_JUDGE_MODEL || "gemini-3.6-flash", id unchanged). judge-video, judge-rubric, find-references and analyze-reference import it; analyze-reference previously embedded the literal in its URL and ignored the env var, so observer and judge could run on different models. Adds a request-shape test (fake fetch through the real scripts) and wires it into npm run check. NODE-LOOPS.md default corrected to gemini-3.6-flash. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Disposition of every runtime site (SPEC §A.2.4)
analyze-reference.mjs:94gemini-3.6-flash(literal in URL, env ignored)JUDGE_MODELfromlib/judge-model.mjs, soGEMINI_JUDGE_MODELapplies to the observer toojudge-video.mjs:183gemini-3.6-flashcalibration/calibration.json,RUBRIC_VERSION); active; now imports the one constantjudge-rubric.mjs:113gemini-3.6-flashfind-references.mjs:61gemini-3.6-flashvoice.mjs:28en_US-lessac-mediumDefer:
generationConfig.temperature 0.2(4 sites; deprecated on 3.6+, ignored today) is removed at the next recalibration with theRUBRIC_VERSIONbump, because it changes judge behaviour; API key in URL query (judge-video.mjs) noted for the same PR. AlsoNODE-LOOPS.md:61default correctedgemini-3.5-flash->gemini-3.6-flash. Override precedence unchanged (GEMINI_JUDGE_MODEL || default).Breaking-parameter fixes
None required now (spec). The new test asserts no
thinking_budget/candidate_countkeys are sent.Test proof (fails before, passes after)
test/judge-model-request.test.mjsruns each real script as a subprocess with a fakefetch(test/support/fake-gemini-fetch.mjs, no network, no key) and asserts the model in the request URL.# tests 9 / pass 7 / fail 2—analyze-reference.mjs: GEMINI_JUDGE_MODEL is honoured(actual URL modelgemini-3.6-flashwhengemini-3.5-flashwas set) andlib/judge-model.mjs exports the one constant(module not found).# tests 9 / pass 9 / fail 0.Check
npm run check(now also runstest:judge-model): exit 0 —parsed 48/48+ JS files, evidence tests 20/20, judge-model tests 9/9,featureclip.evidence-proof/v1 passed: true. The new imports injudge-video.mjs/judge-rubric.mjssit at end of file so.tours/docs/START_HERE.mdline anchors are unchanged.No paid provider calls made; no model eval required (judge id frozen, no id change).
Spec: model-defaults SPEC Part A §A.2 (FeatureClipStudio)
🤖 Generated with Claude Code