Agenta
Agenta is an open-source LLMOps platform that brings prompt management, LLM evaluation, and LLM observability together in one place. Its cloud REST API at https://cloud.agenta.ai/api exposes apps and variants, versioned prompt configurations, evaluations and evaluators, testsets, and OpenTelemetry-based tracing, all authenticated with a Bearer API key. The platform is MIT licensed and can be self-hosted or used via Agenta Cloud.
APIs
Agenta Apps and Variants API
Create and manage LLM applications and their variants, including revisions, forking, committing, and deploying. Variants hold versioned prompt and parameter configurations that ...
Agenta Configs and Prompts API
Fetch and deploy versioned prompt configurations by application, variant, or environment so production code can pull the latest committed prompt without a redeploy. Backs Agenta...
Agenta Evaluations and Evaluators API
Run and manage evaluations and configure evaluators (LLM-as-a-judge, exact match, similarity, custom code, and more) to score variants against testsets, with queues, runs, resul...
Agenta Testsets API
Create, query, version, and import or export the testsets (evaluation datasets) used to drive evaluations, including CSV/JSON file upload and download.
Agenta Observability and Traces API
Ingest LLM telemetry over the OpenTelemetry OTLP/HTTP endpoint and query traces and spans for analytics, sessions, and users. Powers Agenta's LLM observability for cost, latency...
