Skip to main content
Glama

body_local_worker

Process text or images locally to summarize, extract, redact, classify, or describe content; data stays on the machine.

Instructions

Process text (or an image) ON THIS MACHINE with a local model. Nothing leaves.

task: summarise | extract | redact | key | classify | describe text: the text to process. image: base64 PNG/JPEG - only for task='describe' (the eyes for a text-only brain). Take it from eyes_see_desktop(as_image=True) or desktop_watch_read(as_image=True).

This is the private mode's content half: the page text goes to a model running on your machine, so only a short redacted summary ever leaves. It needs Ollama (ollama pull qwen3:0.6b) or a GGUF via BODY_MCP_LOCAL_MODEL - Body-MCP never downloads a model and NEVER falls back to a remote provider: with no local model it says so and stops. 'describe' additionally needs a local VISION model (e.g. ollama pull gemma3:270m): a text-only model would invent a caption, so it refuses instead.

Tool call template

This tool declares no required parameters. Add optional parameters from the input schema as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {},
    "name": "body_local_worker"
  }
}

Input Schema

NameRequiredDescriptionDefault
taskNosummarise
textNo
imageNo

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observedv0.4.0

TDQS

A4.5/5.0
Behavior5/5

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

With no annotations, the description carries the full burden and does so well: it discloses the Ollama/GGUF prerequisite, that Body-MCP never downloads a model, that it NEVER falls back to a remote provider, and that it fails closed ('says so and stops') rather than degrading. The refusal behavior for 'describe' with a text-only model is exactly the kind of failure-mode disclosure an agent needs.

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

Conciseness4/5

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

Front-loaded with the core verb and the privacy guarantee, then requirements, then failure modes in priority order. Some waste and rough formatting (mid-sentence line breaks after 'text-only', inconsistent spacing in the sibling call examples) but no sentence is superfluous.

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

Completeness4/5

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

For a 3-parameter, no-annotation, no-output-schema tool with an external model dependency, the description covers setup, per-task capability limits, parameter sourcing, and failure behavior. The remaining gap is the return shape per task, which is unstated and not covered by any output schema.

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?

Schema coverage is 0% (bare titles and defaults), so the description must compensate, and it does for all three: it enumerates the task values, defines text, and explains image format (base64 PNG/JPEG), its restriction to task='describe', and its source tools. It stops short of saying what each task returns or what 'key' and 'redact' actually produce, leaving those semantics partly to inference.

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?

States a specific verb and resource ('Process text (or an image) ... with a local model') plus the distinguishing property that nothing leaves the machine. The task enumeration (summarise | extract | redact | key | classify | describe) pins down the actual capability set, so an agent can tell this apart from the read/screenshot siblings like eyes_see_desktop.

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?

Gives clear context for the image parameter ('only for task="describe"') and names the two sibling tools that supply images (eyes_see_desktop(as_image=True), desktop_watch_read(as_image=True)). It frames itself as 'the private mode's content half', which implies when to prefer it, but never explicitly states when NOT to use it or what the non-private alternative is.

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