Skip to main content
Glama
Sign In
fm-mcp

fm-mcp

by yrangana
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Rust
Local

Server Configuration

Describes the environment variables required to run the server.

NameRequiredDescriptionDefault
FM_MCP_LOGNoSet to `debug` for more detail. fm-mcp logs to stderr only (stdout carries the MCP protocol).
FM_MCP_REQUEST_TIMEOUT_SECSNoRaise the request timeout limit for long input. Set in the MCP server's environment (default 120).120

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

CapabilityDetails
tools
{}

Tools

Functions exposed to the LLM to take actions

NameDescription
classifyA

Pick the best label for a text from a list you give, with the on-device model for Apple Foundation Models (free, private, offline). Good for triage: sorting tickets, emails, comments or log lines into categories. The answer is always one of your labels (or several, with multi). Input up to about 3,500 words, as text or as path to a text file; 2 to 50 labels. Subtle or ambiguous text can be mislabelled, so spot-check. Not for code, maths or reasoning.

extractA

Extract fields from text into JSON matching a JSON Schema, with the on-device model for Apple Foundation Models (free, private, offline). Good for names, dates, amounts and IDs in short documents, emails, logs. Input up to about 3,500 words, as text or path to a text file. Ask only for fields the text likely states: a missing one is often null but may take a wrong value from nearby text. Flat schemas work best; nested ones are unreliable. Allowed keywords: type, properties, required, items, enum, const, description, minItems, maxItems. Not for code or reasoning; check values that matter.

ocrA

Read the text in an image file (screenshot, scanned page, receipt, handwriting) with the on-device model for Apple Foundation Models (free, private, offline). Give an absolute path to a PNG, JPEG, HEIC, TIFF, GIF or BMP file; PDFs are not supported. Returns the text line by line, or answers prompt about the image instead. Small or unusual text can be misread; check numbers that matter. Not for code review or reasoning.

summariseA

Summarise text with the on-device model for Apple Foundation Models. Free, private and offline. Good for condensing logs, documents, notes and transcripts into a gist (a paragraph, or at most 7 bullets), not a full record. Limit: about 30,000 words of prose but only about 1,000 lines of a dense log; filter bigger logs first (grep the errors). Pass path to a text file instead of text to avoid reading it yourself. Long input is split and combined: slower, and it can drop details. Not for code, maths, reasoning or facts. Summaries can miss or distort details; check anything important.

Prompts

Interactive templates invoked by user choice

NameDescription

No prompts

Resources

Contextual data attached and managed by the client

NameDescription

No resources

Deploy Server
One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

TDQS

A4.5/5.0

Scored across 4 tools

Disambiguation5/5

Each tool targets a clearly distinct operation: classify (pick from given labels), extract (schema-bound fields), ocr (read text from images), and summarise (condense prose). There is no overlap in verb, input type, or output shape, and each description even states what it is NOT for. An agent can select unambiguously.

Naming Consistency4/5

Three tools are lowercase verbs (classify, extract, summarise) in a consistent style with no noun collisions. The single deviation is 'ocr', an acronym/process name rather than a verb, but it is a standard, unambiguous term and the overall set still reads as a predictable pattern.

Tool Count5/5

Four tools is well-scoped for a focused on-device text-processing server. Each tool covers a genuinely different capability (classification, extraction, OCR, summarisation) and none feels redundant or extraneous.

Completeness4/5

The surface covers the main on-device operations: classify, extract, ocr, and summarise, with clear input/output behavior for each. Minor gaps exist, such as no explicit translation, Q&A, or generic generation tool, but the core text-processing lifecycle is well covered and workable.

Maintenance

ActivityMaintained
ResponsivenessSyncing

Resources