Toskar
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| TOSKAR_API_KEY | No | Bearer token required for API/MCP access from other machines when listening beyond loopback. Must be at least 12 characters. Not needed for local loopback use. | |
| TOSKAR_API_HOST | No | Bind address for the API. Set to 0.0.0.0 to allow access from other devices; when listening beyond loopback, TOSKAR_API_KEY is required. | 127.0.0.1 |
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
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| ask_local_aiB | Ask an AI model running on the user's own computer, through Toskar. The question and answer stay on the user's computers. It may read the web or the user's connected knowledge to answer, but it does not change anything. |
| list_local_modelsA | List the AI models Toskar can answer with, including specialized AIs the user trained. |
| search_my_knowledgeA | Search documents, notes, and pages the user connected to Toskar, and return the passages that match. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 3 tools
Each tool has a distinct core action: ask (query the AI), list (enumerate models), search (retrieve passages). There is some overlap because ask_local_ai may internally read connected knowledge, which an agent could confuse with search_my_knowledge, but the descriptions clarify the intended split.
All three tools use a consistent snake_case verb_noun pattern (ask_local_ai, list_local_models, search_my_knowledge), making the set highly predictable.
Three tools is on the lean side but fits the narrow scope of a local AI assistant: query, enumerate models, and search knowledge. Nothing feels redundant, though the surface is minimal.
The core lifecycle (ask, list models, search knowledge) is covered, but there are notable gaps: no way to inspect a specific model's details, list connected knowledge sources, or manage them. Agents can partly work around this via ask_local_ai, but coverage feels thin.