·
2 commits
to main
since this release
🚀 Features
- Align the score
sourceenum withexperiment | annotation | api.SCORE_SOURCE_TYPES(api-schemas) now validates the public score-write API against those three values and defaults an omitted source to"api"(was"eval"); the legacy"eval"value is no longer accepted on write.score()(TS + Python SDK) sendssource, defaulting to"api". Both experiment score-writers stampsource: "experiment": the SDKrunExperimenteval loop and the CLIagentmark run-experimentscore POST. ui-componentsScoreData.sourcewidened to match ("eval"kept only as a legacy display value for historical rows). (#803)
🧱 Updated Dependencies
- Updated @agentmark-ai/api-schemas to 0.9.0
- Updated @agentmark-ai/prompt-core to 1.1.0