Repository navigation
First-draft fallback judge defaults to Claude Sonnet 5.5, not Opus - #109
Merged
Merged
Conversation
A make or refine whose configured judge fails now falls back to the Sonnet engine, so a run never spends Opus usage. Test and TESTING.md assertion updated to the shipped default.
The journey pointed at journeys/fixtures/critic-model.mjs and journeys/fixtures/judge-cli.mjs, which do not exist, so the critic journey could never run.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
What
FALLBACK_JUDGES[0]insrc/judge-fallback.mjswasclaude-opus-5-5. It is nowclaude-sonnet-5-5, so a make or refine run whose configured judge fails falls back to Sonnet and never spends Opus usage.FALLBACK_JUDGES[1](Codexgpt-6.1-sol), the order,DEFAULT_MODELSandjudgeOfare unchanged.Updated because the change made them wrong
src/judge-fallback.mjs: the header comment naming the fallback order.docs/TESTING.md: thecritique.judge_fallbackline naming the fallback judges.test/critic.test.mjs: the assertion on the shipped default model in theJUDGE_ALL_FAILEDerror.test/judge-fallback.test.mjskeepsclaude-opus-5-5; there it is an arbitrary fixture model name, not the shipped default.Fixed because they blocked the proof
journeys/critic.mjs: its two fixture imports pointed atjourneys/fixtures/critic-model.mjsandjourneys/fixtures/judge-cli.mjs, which do not exist in this repo; the real files aretest/fixtures/critic-model.mjsandtest/fixtures/judge-cli.mjs. That made thecriticjourney un-runnable, so it could not produce the proof this PR needs. One path per line, nothing else.Proof
node --import ./test/fixtures/provider-hooks.mjs --test --test-concurrency=1 test/judge-fallback.test.mjs test/critic.test.mjs(the repo's own test command for these files, frompackage.jsontest):Full log:
/home/umer/.treehouse/firstmate-cff959/3/firstmate/data/fd-sonnet-fallback-judge/judge-fallback-critic-tests.logdone-done run criticat commit0906808625f8e0239a5e4228ab35257898ba60ce, clean tree, endedPASS: policy-admitted references, one independent review with seven axes, explicit signed-out refusal(run20261030T111016Z-09068086-74e745).proof: done-done-verify-first-draft (no feature file for this change), done-done run critic at commit 0906808, plus
node --import ./test/fixtures/provider-hooks.mjs --test --test-concurrency=1 test/judge-fallback.test.mjs test/critic.test.mjs(9 pass, 0 fail)Checks
outcome: a make or refine run whose judge fails falls back to Claude Sonnet 5.5 instead of Opus
repro: n/a
proof: done-done-verify-first-draft (no feature file for this change), done-done run critic at commit 0906808, plus
node --import ./test/fixtures/provider-hooks.mjs --test --test-concurrency=1 test/judge-fallback.test.mjs test/critic.test.mjs(9 pass, 0 fail)evidence: 20261030T111016Z-09068086-74e745/refs.json, 20261030T111016Z-09068086-74e745/make.json, 20261030T111016Z-09068086-74e745/refusal.json, 20261030T111016Z-09068086-74e745/steps.json
screen: none