Configure TOMC in WorkBuddy

中文 · Assistant setup · Validation record

This guide provides manual local MCP configuration for Tencent WorkBuddy. WorkBuddy installation and TOMC tool calls have not yet been verified on a real machine. Earlier TOMC server checks keep their own scope. TOMC has no Tencent official marketplace listing or endorsement.

The quickest route, from a checkout with .[mcp] installed (Python 3.10 or newer), is python -m tomc setup --client workbuddy --write, then python -m tomc doctor. The manual route below uses the locked runtime.

Use the locked runtime

Install UV and make uv available on PATH, then clone the public repository:

git clone https://github.com/ZipengWu365/TOMC.git

The repository's plugins/claude-code directory contains the shared TOMC source, pyproject.toml, uv.lock and src/server.py. WorkBuddy can launch that runtime as a local stdio server; installing Claude Code is unnecessary. First startup can download compatible Python and locked dependencies through UV. TOMC itself needs no extra model API key or GPU.

Before connecting, download the locked dependencies with your actual runtime directory:

uv sync --locked --directory "/absolute/path/to/TOMC/plugins/claude-code"

Copy the WorkBuddy template and replace both absolute-path placeholders: the runtime directory and its src/server.py file. Windows JSON paths can use forward slashes, such as C:/Tools/TOMC/plugins/claude-code. If UV is outside PATH, replace command with its actual executable path. Keep the checkout in place and update the two paths after moving it.

Add the server

Open the sidebar 插件 → MCP 服务器 → 配置 MCP (Plugins → MCP servers → Configure MCP), and merge mcpServers.tomc into the JSON while preserving other servers. The official WorkBuddy guide also documents these file locations:

Scope	Configuration file
Personal, across projects	~/.workbuddy/mcp.json
Current project	<project>/.workbuddy/mcp.json

Here ~ means your home directory. After saving, inspect the server status; the guide uses green for a successful connection and red for a configuration or launch error. The template explicitly selects stdio, supported in the official connector reference.

As an alternative, install .[mcp] from TOMC source and run python -m tomc setup --client workbuddy --write. It merges the entry for that Python environment into ~/.workbuddy/mcp.json after saving a timestamped backup, keeps other servers and changes nothing when run again; add --config <project>/.workbuddy/mcp.json for a project file. Without --write it only prints the configuration. Add --store /absolute/path/to/tomc-memory.sqlite3 to select a database. --link remains Cursor-only. After restarting WorkBuddy, check for twelve tools and say Try TOMC Memory.

Tools, storage and feedback

0.3.1 has twelve tools. try_tomc runs a built-in sample. prepare_context prepares supplied history; neither it nor recall_memory saves its result unless you run tomc settings --save-mcp-results on. The notebook tools are remember_memory, recall_memory, list_memories and forget_memory; erase_all_memory deletes everything after confirmation. The status and library tools are get_status, get_usage_stats, list_contexts, get_context and make_handoff. Save only explicitly authorized text and delete only the requested notebook.

Notebooks default to plaintext SQLite at ~/.tomc/memory.sqlite3. For the UV template, merge this field alongside command and args to choose a different file:

{
  "env": {"TOMC_MEMORY_PATH": "/absolute/path/to/tomc-memory.sqlite3"}
}

Clients using the same file can share explicitly saved notes. TOMC does not collect other chats, remove host history or provide cloud synchronization. Returned memory enters WorkBuddy's context and follows its data policy.

Please report setup feedback through GitHub issues, including WorkBuddy version, OS, configuration route and error message. Omit credentials, private history and notebook contents.