Skip to main content
Glama

kb-mcp-server

by Geeksfino
kb.yml2.24 kB
path: .txtai/kb-test # Where to save the index content: true # Store document content writable: true # Enable index writing # Text extraction configuration textractor: paragraphs: true minlength: 50 cleantext: true backend: "docling" # Use Docling for better document parsing with formatting preservation # Embedding configuration embeddings: path: sentence-transformers/nli-mpnet-base-v2 # Model path content: true normalize: true hybrid: true gpu: true # Use GPU if available # Storage configuration writable: true # Vector storage (for embeddings) backend: faiss # Fast vector similarity search # Document storage (for content and metadata) storagetype: sqlite # SQL database for document storage # Scoring methods scoring: method: bm25 normalize: true terms: cachelimit: 1000000000 cutoff: 0.001 # Knowledge graph configuration graph: backend: "networkx" # Graph backend (networkx is recommended) batchsize: 256 # Batch query size for graph building limit: 5 # Reduced from 10 to 5 for more focused results minscore: 0.4 # Minimum similarity score for edges approximate: true # Only run queries for nodes without edges (more efficient) # Topic extraction configuration topics: algorithm: "louvain" # Community detection algorithm (louvain, leiden, etc.) terms: 4 # Number of terms for topic labels # Advanced graph options centrality: "betweenness" # Centrality algorithm (pagerank, betweenness, etc.) directed: true # Whether the graph is directed weight: "similarity" # Edge weight attribute # Graph search options search: max_hops: 1 # Reduced from 3 to 1 for better relevance use_centrality: true # Use centrality for ranking min_score: 0.5 # Increased from 0.3 to 0.5 for better relevance limit: 3 # Limit the number of results to the most relevant ones # Search configuration search: limit: 3 # Limit to top 3 results minscore: 0.5 # Minimum score threshold rerank: true # Enable reranking of results weights: bm25: 0.5 # Weight for keyword matching similarity: 0.5 # Weight for semantic similarity

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/Geeksfino/kb-mcp-server'

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