[BUG] MCP Server Docker image missing optional provider extras (anthropic, google-genai, groq)

[marians]

Bug Description

The MCP server Docker image (zepai/knowledge-graph-mcp:1.0.2-graphiti-0.28.2) installs graphiti-core without optional extras. This means non-OpenAI LLM providers (Anthropic, Gemini, Groq) fail at runtime even though they are listed as supported in the config schema and README.

The MCP server config accepts provider: "anthropic" (and gemini, groq), and the factory code in services/factories.py has code paths for these providers guarded by HAS_ANTHROPIC, HAS_GEMINI, HAS_GROQ flags. But since the extras aren't installed, these flags are False and the providers fail immediately.

Steps to Reproduce

1. Pull the latest MCP server image:

   docker pull zepai/knowledge-graph-mcp:1.0.2-graphiti-0.28.2
   

2. Configure config.yaml with an Anthropic provider:

   llm:
     provider: "anthropic"
     model: "claude-haiku-4-5-20251001"
     temperature: 0.0
     providers:
       anthropic:
         api_key: "${ANTHROPIC_API_KEY}"

3. Start the server. It fails during initialization.

Expected Behavior

The Docker image should include optional extras for all providers that the MCP server advertises as supported, so that switching providers is a config-only change.

Actual Behavior

WARNING - Failed to create LLM client: Anthropic client not available in current graphiti-core version

The HAS_ANTHROPIC flag in factories.py is False because graphiti-core[anthropic] was not installed. Same applies to HAS_GEMINI and HAS_GROQ.

Environment

• Graphiti Version: graphiti-core 0.28.2, MCP server 1.0.2 (Docker image zepai/knowledge-graph-mcp:1.0.2-graphiti-0.28.2)
• Python Version: 3.11
• Operating System: Docker (Debian-based)
• Database Backend: FalkorDB
• LLM Provider & Model: Anthropic claude-haiku-4-5-20251001

Installation Method

• Docker (zepai/knowledge-graph-mcp:1.0.2-graphiti-0.28.2)

Workaround

Build a custom image that installs the extras. The image uses uv:

FROM zepai/knowledge-graph-mcp:1.0.2-graphiti-0.28.2
RUN cd /app/mcp && uv add "graphiti-core[anthropic,google-genai]==0.28.2"

Possible Solution

Change the Dockerfile for zepai/knowledge-graph-mcp to install graphiti-core with all optional provider extras:

graphiti-core[anthropic,google-genai,groq]==0.28.2

The additional dependencies are lightweight (the anthropic, google-genai, and groq Python packages) and would make the image work out of the box for all supported providers.

Related

• [BUG] MCP Server: cross_encoder/reranker not configurable, hardcoded to OpenAI — requires OPENAI_API_KEY even with non-OpenAI providers #1393 — MCP server also doesn't configure the cross-encoder for non-OpenAI providers
• Question: How to setup the graphiti to use my OpenRouter and Voyage? #517 — User confusion about configuring non-OpenAI providers in MCP Docker setup