Skip to main content
Glama

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
OLLAMA_HOSTNoOllama server URL for Ollama embeddings and LLMhttp://localhost:11434
GEMINI_API_KEYNoGoogle Generative AI key for Gemini embedding and Gemini Flash LLM
MNEMOSTACK_LLMNoLLM provider for answer generation and reranking
MNEMOSTACK_PROVIDERNoEmbedding provider for CLI/HTTP/MCP
MNEMOSTACK_GRAPH_URINoMemgraph bolt URI for graph retriever and GraphStore
MNEMOSTACK_BM25_PATHSNoBM25 corpus paths separated by os.pathsep (':' on Unix) for BM25 retriever
MNEMOSTACK_COLLECTIONNoQdrant collection namemnemostack
MNEMOSTACK_QDRANT_URLNoQdrant server URL for remote Qdranthttp://localhost:6333
MNEMOSTACK_LLM_PROVIDERNoAlternative LLM provider (alias for MNEMOSTACK_LLM)
MNEMOSTACK_MEMGRAPH_URINoAlternative Memgraph bolt URI (alias for MNEMOSTACK_GRAPH_URI)
MNEMOSTACK_AUTO_RECORD_IORNoToggle for HTTP recall exposure logging (true/false)
MNEMOSTACK_EMBEDDING_PROVIDERNoAlternative embedding provider (alias for MNEMOSTACK_PROVIDER)

Capabilities

Features and capabilities supported by this server

CapabilityDetails
tools
{
  "listChanged": true
}
logging
{}
prompts
{
  "listChanged": false
}
resources
{
  "subscribe": false,
  "listChanged": false
}
extensions
{
  "io.modelcontextprotocol/ui": {}
}
experimental
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
mnemostack_healthB

Check health of mnemostack components (embedding, vector store, optional graph).

mnemostack_searchA

Hybrid recall over indexed memories.

Returns top-K results ranked by reciprocal rank fusion of BM25 and semantic search. Each result has id, text, score, sources, payload.

mnemostack_answerB

Generate concise factual answer from retrieved memories.

Uses hybrid recall to find relevant memories, then an LLM inference layer to synthesize a short answer with confidence score and citations.

Returns: answer text, confidence (0.0-1.0), sources, fallback_recommended.

mnemostack_feedbackA

Record explicit feedback for stateful recall learning.

Use signal='clicked' to also record inhibition-of-return exposure. Pass retriever labels from mnemostack_search results as sources so Q-learning can update source weights.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

Latest Blog Posts

MCP directory API

We provide all the information about MCP servers via our MCP API.

curl -X GET 'https://glama.ai/api/mcp/v1/servers/udjin-labs/mnemostack'

If you have feedback or need assistance with the MCP directory API, please join our Discord server