fix(schema): reject stale authored skill and trajectory graders#1686
Merged
Conversation
Deploying agentv with
|
| Latest commit: |
12232f6
|
| Status: | ✅ Deploy successful! |
| Preview URL: | https://547a2fa7.agentv.pages.dev |
| Branch Preview URL: | https://promptfoo-stale-grader-depre.agentv.pages.dev |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Authored eval YAML now hard-rejects the stale AgentV-only
skill-triggerandtool-trajectoryassertion types with migration guidance to the Promptfoo-compatibleskill-used,not-skill-used, andtrajectory:*shapes.The migration script can rewrite the supported stale forms, while latency-specific
tool-trajectorychecks are left for explicit validation failure because there is no Promptfoo-compatible equivalent yet. Existing examples and AI-facing skill-data guidance now teach the replacement assertions, while private runtime compatibility types/tests remain in place.Validation
bun test scripts/migrate-hard-deprecations.test.ts packages/core/test/evaluation/loaders/grader-parser.test.ts packages/core/test/evaluation/validation/eval-file-schema.test.ts packages/core/test/evaluation/validation/eval-validator.test.ts packages/core/test/evaluation/validation/eval-schema-sync.test.tsbun run typecheckbun run lintbun run validate:examplesgit diff --checkLive dogfood was not run because this PR changes authored YAML validation/parsing, codemod migration, examples, and guidance only. It does not change eval execution or grader runtime behavior beyond rejecting stale authored config before runtime.
Post-Deploy Monitoring & Validation
agentv validateagainst migrated downstream evals that previously authoredskill-triggerortool-trajectory.Related
Related: av-kfik.45.1