Skip to main content
Glama
Local AI MCP

Local AI MCP

Official
by TMHSDigital

Pull model

pull_model
Idempotent

Download a model onto a chosen local AI provider, with progress streaming or background job support for long-running pulls.

Instructions

HEAVY: Download a model onto ONE provider (multiple GB, can take a long time). Streams download progress via MCP progress notifications when the client sends a progressToken. With background:true, returns a jobId immediately instead; poll pull_status and stop it with pull_cancel (use this when the client has no progress display or a short tool timeout). Without provider, uses the provider that already has the model, else the first live local provider that can pull.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "model": "<model>"
    },
    "name": "pull_model"
  }
}

Input Schema

NameRequiredDescriptionDefault
modelYesModel id/name to pull (e.g. qwen3:8b, hf.co/<repo>:<quant>)
providerNoProvider instance name. Omit to let the server choose (local providers only, unless LOCAL_AI_ALLOW_HOSTED_FALLBACK is set).
backgroundNoReturn a jobId at once and download in the background (default false). A second call for the same model joins the running job.

Output Schema

NameRequiredDescriptionDefault
jobNo
pullNo

Schema Changelog

Changes observed during successful MCP inspections.

  1. Changed3 schema fields changedv0.6.1
    • addedInput schema / properties / background
      Added value: +{
      +  "description": "Return a jobId at once and download in the background (default false). A second call for the same model joins the running job.",
      +  "type": "boolean"
      +}
    • addedOutput schema / properties / job
      Added value: +{
      +  "additionalProperties": {},
      +  "propertyNames": {
      +    "type": "string"
      +  },
      +  "type": "object"
      +}
    • removedOutput schema / required
      Removed value: -[
      -  "pull"
      -]
  2. Changed5 schema fields changedv0.5.0
    • removedInput schema / additionalProperties
      Removed value: -false
    • changedInput schema / properties / model / description
      Previous value: -"Model id/name to pull (download)"New value: +"Model id/name to pull (e.g. qwen3:8b, hf.co/<repo>:<quant>)"
    • changedInput schema / properties / provider / description
      Previous value: -"Optional provider id"New value: +"Provider instance name. Omit to let the server choose (local providers only, unless LOCAL_AI_ALLOW_HOSTED_FALLBACK is set)."
    • changedInput schema / properties / provider / enum
      Previous value: -[
      -  "ollama",
      -  "lmstudio",
      -  "moonshot",
      -  "openaicompat",
      -  "llamacpp"
      -]New value: +[
      +  "ollama",
      +  "lmstudio",
      +  "llamacpp",
      +  "dockermodelrunner"
      +]
    • changedOutput schema / (root)
      Previous value: -nullNew value: +{
      +  "$schema": "http://json-schema.org/draft-07/schema#",
      +  "additionalProperties": false,
      +  "properties": {
      +    "pull": {
      +      "additionalProperties": {},
      +      "propertyNames": {
      +        "type": "string"
      +      },
      +      "type": "object"
      +    }
      +  },
      +  "required": [
      +    "pull"
      +  ],
      +  "type": "object"
      +}
  3. First observedv0.4.0

TDQS

A4.7/5.0
Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

The description carries substantial behavior beyond the annotations: multi-GB size, long duration, streaming via MCP progress notifications gated on a progressToken from the client, an immediate jobId under background:true, and provider selection fallback (already-has-model, else first live local provider, with LOCAL_AI_ALLOW_HOSTED_FALLBACK as the hosted escape hatch). The "a second call for the same model joins the running job" note gives concrete meaning to idempotentHint:true, and nothing conflicts with the annotations.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

The cost warning is front-loaded as the first token ("HEAVY:"), then progress mode, then background mode, then provider resolution — an efficient ordering with no filler sentences. Every clause either sets expectations or changes how the agent should call it.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a heavyweight, long-running, job-capable tool this covers the operational facts an agent needs: cost, progress mechanics, async path and its polling/cancel siblings, and provider resolution defaults. An output schema is declared, so return-value documentation is correctly omitted from the description.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

With 100% schema coverage the schema already explains all three parameters, so this starts at baseline 3. The description adds genuine semantics the schema does not: what happens when provider is omitted (existing-model preference, then first live local provider, hosted only under an env flag) and the deduplication behavior of repeated background calls.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

"HEAVY: Download a model onto ONE provider" gives a specific verb and resource plus an explicit scoping constraint (single provider), and the surrounding text carves it apart from pull_status and pull_cancel by role (this one starts the download; those poll and stop it). An agent can tell immediately that this is the initiating operation rather than a status or management call.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

It gives a concrete decision rule for the background mode ("use this when the client has no progress display or a short tool timeout") and names the follow-up tools to poll and cancel, which is real when-to-use guidance. It does not address when to prefer ensure_model over an explicit pull, so one obvious sibling comparison is left to inference.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.