RagaAI

license

Apache-2.0(covers only the ragaai-catalyst client library, not the platform it talks to)

source

closed(the observability/eval platform (project, dataset, trace and prompt storage, evaluation execution) runs at catalyst.raga.ai

platform

proprietary-SaaS(every class in the README requires an access_key/secret_key issued by the hosted platform)

self-host

no(no server, dashboard or docker-compose file anywhere in the repository)

RagaAI Catalyst's repository is a full-featured client SDK for project, dataset, evaluation, trace, prompt and guardrail management, but every call authenticates against the hosted catalyst.raga.ai service, and none of that platform is published. With no server, dashboard or self-host path in the repository, this is an open client over a closed hosted product.

• https://api.github.com/repos/raga-ai-hub/RagaAI-Catalyst/git/trees/main?recursive=1 recorded 2026-10-06

  The whole repository tree, 284 paths, truncated:false. Searched for docker, compose, self-host, server, backend and dashboard paths: none exist. The only license-related path is the root LICENSE file (Apache-2.0).

• https://raw.githubusercontent.com/raga-ai-hub/RagaAI-Catalyst/main/LICENSE recorded 2026-10-07

  Apache License, Version 2.0 full text, the sole license file in the repository. It covers this client library, not the hosted platform it authenticates against.

• https://raw.githubusercontent.com/raga-ai-hub/RagaAI-Catalyst/main/README.md recorded 2026-10-07

  The Configuration section: "Before using RagaAI Catalyst, you need to set up your credentials" via `access_key`, `secret_key` and `base_url`, with the note "Authentication to RagaAICatalyst is necessary to perform any operations below." Every subsequent usage example (Project, Dataset, Evaluation, Trace and Prompt Management) instantiates against that authenticated client.

No usage figure exists for the hosted Catalyst platform: the ragaai-catalyst PyPI package and its GitHub repository are the client SDK, so their downloads and stars measure that client rather than the platform, and the vendor's site names no customers or usage figure for Catalyst. Rather than infer adoption from the client's numbers, the map leaves this product unscored.

• https://api.github.com/repos/raga-ai-hub/RagaAI-Catalyst recorded 2026-09-02

  stargazers_count = 16,157, forks_count = 3,569, pushed_at 2026-02-11, for the client SDK repository.

• https://pypistats.org/api/packages/ragaai-catalyst/recent recorded 2026-09-02

  last_month = 2,711 downloads for ragaai-catalyst (last_week 359, last_day 236). A measurement of the client SDK's channel, recorded as context and explicitly not the basis of a reading.

• https://raga.ai/ recorded 2026-09-02

  A 3 KB HTML shell for a client-rendered app. The only readable text is the title and meta description - "RagaAI | Enterprise Healthcare AI Agent Suites", "purpose-built, production-grade AI Agent Suites for Healthcare, Lifesciences, and Aerospace. Powered by Prism and Catalyst" - with no customer roster, user count or download figure for Catalyst anywhere in the served body.

RagaAI Catalyst covers every dimension this category grades - tracing, evaluation, prompt and dataset management - plus synthetic data generation, guardrails and red-teaming that few peers in this category offer. That breadth keeps it level with Langfuse, even though none of it, unlike Langfuse, is self-hostable.

• https://raw.githubusercontent.com/raga-ai-hub/RagaAI-Catalyst/main/README.md recorded 2026-10-07

  Table of contents and usage sections for Project Management, Dataset Management, Evaluation, Trace Management, Agentic Tracing, Prompt Management, Synthetic Data Generation, Guardrail Management and Red-teaming, each with a working code example.