Compatibility
What we have tested.
Each cell comes from a test run, with its date. Anything without a run says Untested: it may work, but we have not checked it. CI rebuilds this page from its runs; last update 2026-10-19.
No run has reported yet, so every cell below is Untested.
Frameworks
Each framework runs its guide: it connects with the framework's own MCP client, saves a memory with remember and finds it with recall. The first column runs against a local copy of the MCP server on every change; the second runs against agentics.you daily.
| Framework | Tested version | Test server | agentics.you |
|---|---|---|---|
| LangChain | langchain-mcp-adapters 0.3.2 | Untested | Untested |
| LlamaIndex | llama-index-tools-mcp 0.4.8 | Untested | Untested |
| CrewAI | crewai 1.15.28 | Untested | Untested |
| OpenAI Agents SDK | openai-agents 0.24.0 | Untested | Untested |
| Vercel AI SDK | ai 7.0.143 | Untested | Untested |
AI clients
A client passes when someone connected it with its own sign-in and the launch probe saved and recalled a memory on agentics.you with that connection.
Models
The guides hand the same tools to an agent that runs on a model. CI checks that step’s code loads, but no run has called a model yet.
| Model | LangChain | LlamaIndex | CrewAI | OpenAI Agents SDK | Vercel AI SDK |
|---|---|---|---|---|---|
| OpenAI gpt-4.1 | Untested | Untested | Untested | Not used by this guide | Untested |