AgentOps

license

Elastic-License-2.0(app/LICENSE, the AgentOps application)+MIT(OSI, root LICENSE, the agentops SDK)

source

public(SDK + self-hostable app dir)

commercial

app.agentops.ai-SaaS-dashboard-is-the-primary-product

core-gated

ungated(app/ ships the whole platform - api, dashboard, clickhouse, opentelemetry-collector, supabase, deploy, landing and compose.yaml - with no ee/ or enterprise directory, no license-key check and no premium flag)

The score is taken from app/LICENSE, which is the Elastic License 2.0 and forbids providing the software to third parties as a hosted or managed service. That is a restriction on competing with the vendor, and with the source published it lands on the source-available rung at 2. The repository root carries plain MIT with no carve-out, but that governs the root, where the agentops SDK lives, and says nothing about app/; it describes the SDK rather than the observability application scored here. Both licenses are recorded together so that the more restrictive one decides rather than only the first being read. Nothing is withheld for a paid tier - app/ ships the API, dashboard, ClickHouse, the OpenTelemetry collector, Supabase, deploy tooling and a compose file, with no enterprise directory and no license key - but a license that restricts competition outranks an ungated core. The vendor's README calls the app MIT; app/LICENSE says Elastic 2.0, and the license file governs.

• https://github.com/AgentOps-AI/agentops recorded 2026-06-04

  MIT-licensed SDK; complementary SaaS dashboard at app.agentops.ai; self-host option

• https://raw.githubusercontent.com/AgentOps-AI/agentops/main/app/LICENSE recorded 2026-08-12

  The license on the AgentOps app, which is not the license this record carries. The file is the Elastic License 2.0 verbatim. Its Limitations: "You may not provide the software to third parties as a hosted or managed service, where the service provides users with access to any substantial set of the features or functionality of the software", and "You may not move, change, disable, or circumvent the license key functionality in the software, and you may not remove or obscure any functionality in the software that is protected by the license key." Elastic-License-2.0 is declared in software.yaml's competition_restricted examples. The only commit touching this path is "AgentOps OSS release (#1190)", 2025-08-09, so it is the license the app was published under rather than a later change.

• https://raw.githubusercontent.com/AgentOps-AI/agentops/main/LICENSE recorded 2026-08-12

  The root LICENSE, which is plain MIT with no carve-out clause of any kind - unlike langfuse, agenta or litellm, it does not name a directory it excludes. It therefore governs the repository root, where the agentops Python SDK lives, and says nothing about app/, which carries its own LICENSE. GitHub's license endpoint reports MIT for the repo on the strength of this file.

• https://api.github.com/repos/AgentOps-AI/agentops/contents/app recorded 2026-08-12

  The app directory, listed to test whether anything is withheld. It holds LICENSE, README.md, SETUP_GUIDE.md, api, dashboard, clickhouse, opentelemetry-collector, deploy, e2e, graphana, landing, supabase, scripts and compose.yaml. There is no ee/, no enterprise/ and no second license below it. The published tree is the running platform, so on the core_gated question nothing is held back.

• https://raw.githubusercontent.com/AgentOps-AI/agentops/main/README.md recorded 2026-08-12

  The vendor's own claims, recorded including the one the LICENSE contradicts. Under "Open Source": "The AgentOps app is open source under the MIT license. Explore the code in our app directory." Under "Self-Hosting": "Looking to run the full AgentOps app (Dashboard + API backend) on your machine? Follow the setup guide in app/README.md", and the feature table lists "Self-Host: Want to run AgentOps on your own cloud? You're covered". The self-host claim checks out against app/; the MIT claim does not check out against app/LICENSE.

agentops draws 275,111 PyPI downloads a month, inside the 100K-1M band. It integrates with CrewAI, AG2, the OpenAI Agents SDK, LangChain and Camel, and roughly 5,600 GitHub stars corroborate without setting the band. Downloads have fallen from a high of 584,603, which is consistent with the stalled release cadence recorded on the capability axis.

• https://pypistats.org/packages/agentops recorded 2026-08-13

  Downloads last month 275,111 for agentops (last week 59,009)

• https://github.com/AgentOps-AI/agentops recorded 2026-08-13

  5,771 stars; agent tracing, cost tracking and framework integrations in the README

Solid agent execution tracing with session replays, LLM cost tracking, multi-framework integrations, monitoring and debugging dashboards and custom evaluation metrics, but a narrower feature matrix than MLflow, Langfuse, Opik or Phoenix: prompt versioning and dataset or annotation workflows are thin, and evaluations are carried as a roadmap table rather than a shipped surface. A stale SDK with no release for over a year reinforces the mid score, one rung below Langfuse at 4.

• https://github.com/AgentOps-AI/agentops recorded 2026-08-13

  Repository README - "AgentOps helps developers build, evaluate, and monitor AI agents", with a feature table covering step-by-step agent execution graphs, LLM cost management, framework integrations and Self-Host, and an "Evaluations Roadmap" table in which several eval rows are still marked as in-progress rather than shipped.