BREAKING: consolidate 'metrics' and 'custom_evaluators' into evaluators#149
Merged
Conversation
…s field Signed-off-by: Peter Jausovec <peter.jausovec@solo.io>
evaluators
evaluatorsevaluators
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Warning
Breaking change. The eval config shape has changed. Existing
YAML config files, JSON API payloads, and MCP
evaluate-sessionsrequests need migration. See the migration guide below.
Previously we had a list of strings (
metrics) and separatecustom_evaluatorsfield. this meant that built-in metrics would be provided through themetricsvalue as a name, and only run-level configuration would be applied -- i.e. there was not way to configure per-builtin-evaluator metrics.The evaluation config now requires
evaluatorsandagentevalsnow accepts a single canonical evaluation config field:evaluators.The legacy
metricsfield has been removed, and legacy top-level config shapes are no longer accepted. Requests and config files must now define built-in and custom evaluators uniformly underevaluators[].Before:
{ "metrics": ["tool_trajectory_avg_score"] }After:
{ "evaluators": [ { "name": "tool_trajectory_avg_score", "type": "builtin" } ] }This also means older legacy config fields such as customEvaluators and top-level builtin overrides like judgeModel, threshold, and trajectoryMatchType must be migrated into the corresponding evaluator entries inside
evaluators[].Migration
YAML eval config files
Before:
After:
/api/evaluate,/api/evaluate/stream,/api/evaluate/json,/api/runsBefore:
{ "metrics": ["tool_trajectory_avg_score"], "judgeModel": "gemini-2.5-flash" }After:
{ "evaluators": [ { "name": "tool_trajectory_avg_score", "type": "builtin", "judgeModel": "gemini-2.5-flash" } ] }Legacy keys now return 4xx with
Extra inputs are not permitted.MCP
/api/streaming/evaluate-sessionsSame shape as
/api/evaluateabove. The oldmetrics/judge_model/
trajectory_match_typetop-level fields are rejected.CLI
agentevals list-metricshas been removed. Useagentevals evaluator list --source builtininstead.