Confluent Support RAG MCP Server
by BhwinGT
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| MCP_HOST | No | MCP server bind host. In Docker, overridden to 0.0.0.0. | 127.0.0.1 |
| MCP_PORT | No | MCP server port. | 8000 |
| CHUNK_SIZE | No | Target words per chunk. | 300 |
| EMBED_MODEL | No | Dense embeddings model. Always local. | qwen3-embedding:0.6b |
| OLLAMA_HOST | No | Ollama host. In Docker, overridden to http://ollama:11434. | http://127.0.0.1:11434 |
| QDRANT_HOST | No | Qdrant host. In Docker, overridden to qdrant. | localhost |
| QDRANT_PORT | No | Qdrant port. | 6333 |
| NGROK_DOMAIN | No | Reserved ngrok domain (hostname only). | |
| TOP_K_RESULTS | No | Default number of results per search. | 5 |
| ALLOWED_EMAILS | No | Comma-separated allowed exact emails (auth mode B, overrides ALLOWED_EMAIL_DOMAINS). | |
| OLLAMA_API_KEY | No | Set to use Ollama Cloud for chat in orchestrator.py. | |
| PUBLIC_MCP_URL | No | Public HTTPS URL. Must match Entra redirect URI. Required when MCP_AUTH_ENABLED=true. | |
| SUPPORT_COOKIE | No | Browser session cookie from logged-in Confluent support portal. Needed for scraping. | |
| AZURE_API_SCOPE | No | API scope exposed in Entra. | access |
| AZURE_CLIENT_ID | No | Entra app client ID. Required when MCP_AUTH_ENABLED=true. | |
| AZURE_TENANT_ID | No | Entra tenant ID. Required when MCP_AUTH_ENABLED=true. | |
| INFERENCE_MODEL | No | Model used by orchestrator.py for chat. | gemma4:31b |
| NGROK_AUTHTOKEN | No | ngrok authtoken for remote access. | |
| MCP_AUTH_ENABLED | No | Turns on Microsoft sign-in. Set to 'true' to enable. | false |
| RERANK_MIN_SCORE | No | Raw cross-encoder score threshold. Negatives are normal. | 0 |
| SUPPORT_BASE_URL | No | Base URL for scraping support articles. | https://support.confluent.io |
| QDRANT_COLLECTION | No | Qdrant collection name. | confluent_support |
| AZURE_TOKEN_ISSUER | No | Only if the app still issues v1.0 tokens. | |
| AZURE_CLIENT_SECRET | No | Entra app client secret. Required when MCP_AUTH_ENABLED=true. | |
| FASTEMBED_CACHE_DIR | No | Cache directory for fastembed models. | ./model_cache |
| MCP_JWT_SIGNING_KEY | No | Signing key for session tokens. Set to persist sessions across restarts. | |
| SUPPORT_CATEGORY_IDS | No | Comma-separated Zendesk category IDs. | 360004404872,360005831891,360005822052 |
| ALLOWED_EMAIL_DOMAINS | No | Comma-separated allowed email domains (auth mode A). | |
| CLOUDFLARE_TUNNEL_TOKEN | No | Cloudflare Tunnel token (alternative to ngrok). |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Server capabilities have not been inspected yet.
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
No tools | |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
This server cannot be deployed
Maintenance
ActivitySlowing
ResponsivenessNo issues