ThreadShelf MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| HOST | No | Server host. Set explicitly only when trusted LAN access is required. | 127.0.0.1 |
| PORT | No | Server port. | 3000 |
| UPLOADS_DIR | No | Uploaded source files. | .uploads |
| LANCEDB_PATH | No | LanceDB directory. | .lancedb |
| ALLOWED_HOSTS | No | Comma-separated extra Host/Origin names for trusted LAN access. | |
| CHUNK_MAX_CHARS | No | Max characters per embedded chunk. | 2000 |
| COLLECTIONS_PATH | No | Manual-collections registry file. | .collections.json |
| EMBED_BATCH_SIZE | No | Embedding batch size during ingest. | 25 |
| LLAMA_CPP_SERVER | No | Absolute path to an existing llama-server executable. | |
| LLAMA_CPP_THREADS | No | CPU generation threads; -1 lets llama.cpp choose. | -1 |
| LLAMA_MODEL_PATHS | No | Extra model roots (; on Windows, : on macOS/Linux). | |
| LLAMA_CPP_BASE_URL | No | Existing loopback-only llama.cpp server URL. | |
| LLAMA_CPP_MAIN_GPU | No | Main GPU index for applicable profiles. | 0 |
| OPENROUTER_API_KEY | No | OpenRouter key; may be set in .env, never exposed to the browser. | |
| CHUNK_OVERLAP_CHARS | No | Overlap between long chunks. | 100 |
| MASTER_PROMPTS_PATH | No | Saved master (system) prompts. | .threadshelf/master-prompts.json |
| OPENROUTER_BASE_URL | No | Override primarily intended for testing. | https://openrouter.ai/api/v1 |
| LLAMA_CPP_GPU_LAYERS | No | Exact layer offload for the hybrid profile. | 20 |
| LLAMA_CPP_SPLIT_MODE | No | Multi-GPU split: layer or row. | layer |
| GENERATION_CONFIG_PATH | No | Non-secret Experimental Alpha generation settings. | .threadshelf/generation.json |
| LLAMA_CPP_ACCELERATION | No | auto, cpu, gpu, hybrid, or multi-gpu. | auto |
| LLAMA_CPP_CONTEXT_SIZE | No | Managed local server context size. | 8192 |
| LLAMA_CPP_TENSOR_SPLIT | No | Optional multi-GPU proportions, e.g. 3,1. | |
| THREADSHELF_TOOLS_PATH | No | llama.cpp discovery/installer root. | .threadshelf/tools |
| LLAMA_CPP_FLASH_ATTENTION | No | Flash Attention: auto, on, or off. | auto |
| THREADSHELF_DISABLE_DEFAULT_MODEL_PATHS | No | Set 1 to scan only explicitly configured roots. | 0 |
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
ActivityMaintained
ResponsivenessUnresponsive