Convert agent-041's blocking-data check to the agentic LLM judge#95630
Merged
Conversation
The whole-file /getDashboardData\s*\(/ ban rejected functionally correct streaming (async section components defined in page.tsx, or the documented start-promise-without-await pattern) while a byte-identical solution split across two files passed. Follows the agent-034/030 judge conversions.
Contributor
Stats from current PR✅ No significant changes detected📊 All Metrics📖 Metrics GlossaryDev Server Metrics:
Build Metrics:
Change Thresholds:
⚡ Dev Server
📦 Dev Server (Webpack) (Legacy)📦 Dev Server (Webpack)
⚡ Production Builds
📦 Production Builds (Webpack) (Legacy)📦 Production Builds (Webpack)
📦 Bundle SizesBundle Sizes⚡ TurbopackClient Main Bundles
Server Middleware
Build DetailsBuild Manifests
📦 WebpackClient Main Bundles
Polyfills
Pages
Server Edge SSR
Middleware
Build DetailsBuild Manifests
Build Cache
🔄 Shared (bundler-independent)Runtimes
📎 Tarball URLCommit: fc20f46 |
Contributor
Tests PassedCommit: fc20f46 |
…uild) Same accommodation agent-030/034 got with their judge conversions: next build type-checks the workspace and cannot resolve @vercel/agent-eval/eval, failing every run at the build step even when all eval assertions pass.
aurorascharff
approved these changes
Jul 9, 2026
gaojude
added a commit
to vercel/next-evals-oss
that referenced
this pull request
Jul 9, 2026
Sync pin moves to vercel/next.js#95630 (agent-041's blocking-data check becomes an agentic judge; its old whole-file regex failed functionally correct streaming). Rerun agent-041 for claude-fable-5, gpt-5.6-xhigh and gpt-5.6-sol-xhigh (base + agents-md): all six pass. Sol base now ties gpt-5.6 at 88% — its previous gap was entirely the regex artifact. Also clears claude-fable-5--agents-md's 030/034/040 backlog (030 and 034 pass, 040 is a genuine 0/4) and re-records sol's agent-031 with its missing classification. gpt-5.6-xhigh--agents-md needed timeout 2400 (with the docs pointer it reasons 1500-2200s on this eval); cached results carried forward via refingerprint. ACCEPTED_STALE regenerated from live status — remaining models keep their stale agent-041 (exported as notAvailable) until rerun.
gaojude
added a commit
to vercel/next-evals-oss
that referenced
this pull request
Jul 9, 2026
gaojude
added a commit
to vercel/next-evals-oss
that referenced
this pull request
Jul 9, 2026
…ged evals (#97) * Adopt the agent-041 judge conversion for fable and both GPT 5.6 models Sync pin moves to vercel/next.js#95630 (agent-041's blocking-data check becomes an agentic judge; its old whole-file regex failed functionally correct streaming). Rerun agent-041 for claude-fable-5, gpt-5.6-xhigh and gpt-5.6-sol-xhigh (base + agents-md): all six pass. Sol base now ties gpt-5.6 at 88% — its previous gap was entirely the regex artifact. Also clears claude-fable-5--agents-md's 030/034/040 backlog (030 and 034 pass, 040 is a genuine 0/4) and re-records sol's agent-031 with its missing classification. gpt-5.6-xhigh--agents-md needed timeout 2400 (with the docs pointer it reasons 1500-2200s on this eval); cached results carried forward via refingerprint. ACCEPTED_STALE regenerated from live status — remaining models keep their stale agent-041 (exported as notAvailable) until rerun. * Keep last measurements for changed evals; only never-run evals export as N/A A changed eval (judge conversion, content tweak) keeps the model's last measured result on the board until that model is rerun — previously it exported as notAvailable, punching holes in every non-rerun model's score during incremental adoption. Never-run evals (new or renamed upstream, e.g. agent-040-instant for pre-rename models) still export as notAvailable and count against the rate. grok-4.5's agent-041 rerun is blocked upstream: the AI Gateway now returns 'Your team has restricted access to this model' for xai/grok-4.5 (worked 2026-07-08, verified restricted 2026-07-09). Its stale 041 pass stays on the board under this policy; pair remains in ACCEPTED_STALE for a rerun once access is restored. * Re-pin to the merged vercel/next.js#95630 canary commit
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
2 participants
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.
Make the ppr test accept diverse solutions as long as they lead to a static shell. Achieved by
expect(environment).toSatisfyCriterion(a.k.a., llm-as-judge)