ReMCP
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
No arguments | |||
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": true
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| read_fileA | Read a plain text or line-oriented file on this computer: source, config, Markdown, scripts or logs. Use read_document for PDF, DOCX, XLSX and other structured documents. Legacy .docx/.pdf extraction remains supported for compatibility. Use offset and length to page through large text files; a negative offset reads from the end. |
| manage_filesA | Use this when the task is about files, directories, archives, or filesystem search on this computer. Choose exactly one operation; parameters must contain only that operation’s arguments and are validated against its closed schema. Write, edit, move, permission, and delete operations can change or remove local data, while read and search operations leave files unchanged. Use read_file for one known text file. |
| run_terminalA | Use this when the task needs a shell command or an existing ReMCP terminal session: start, read, wait, interact, stop, or list sessions. Choose exactly one operation; parameters are validated against that operation’s closed schema. Starting or interacting with commands can change local or external state, and force termination can lose unsaved process work; reading, waiting, and listing do not mutate the session. |
| control_computerA | Use this for semantic inspection or control of native desktop applications on this computer: windows, accessibility/UI elements, waits, displays, app launch, text/input, clipboard, open/reveal, notifications, or short screen recording. Snapshot/find/wait operations observe state; clicks, typing, window close, app launch, clipboard writes, and recording can change local application state. Prefer semantic targets before raw coordinates. |
| view_imageA | Use this when the result must be pixels from this computer: read an existing image or capture the desktop, one window, monitor, or region. These operations return pixels without interacting with the visible UI. Use control_computer when semantic structure or an action is enough. |
| manage_systemA | Use this for operating-system or ReMCP runtime state: system/runtime/process facts, runtime preferences, services, event logs, networking, installed apps, environment, audio, or explicit power actions. Fact, log, and inventory operations only observe state; process termination, preference/service/audio changes, restart, and shutdown can interrupt work or change the machine. Use run_terminal for general shell commands. |
| manage_documentsA | Use this for structured document content: read PDF/DOCX/XLSX, edit DOCX/XLSX directly, or merge, split, extract, or inspect PDFs. DOCX/XLSX edits can replace the input when no separate output is supplied, and PDF write operations create or replace their requested output; read_document itself is read-only. Use read_file for plain line-oriented text. |
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 7 tools
Each tool description explicitly routes away from its nearest neighbors (read_file vs manage_documents, view_image vs control_computer, run_terminal vs manage_system), which is unusually good. However, real overlaps remain: read_file admits it still does 'legacy .docx/.pdf extraction', manage_files has read/search operations, and control_computer's screen recording overlaps view_image's capture, so an agent can still misselect in those seams.
All names are snake_case verb_noun and readable, but the verb set is mixed (manage_*, run_, control_, read_, view_), giving four different verb families across seven tools. The three manage_* siblings form a clean group while the remaining four each use a different verb, a minor inconsistency rather than a real problem.
Seven tools is well within the ideal range and each maps to a coherent capability domain (terminal, OS, GUI, documents, files, text, pixels). The multi-operation design is deliberate, with each tool exposing a closed operation schema rather than spawning dozens of narrow tools.
The surface covers shell, system/runtime, desktop GUI, office documents, filesystem, plain text, and pixels — essentially the full native-computer-control lifecycle, including writes, edits, and captures. Minor gaps remain: no explicit HTTP/web or network fetch capability, and image writing/encoding beyond capture is not evident.