One-click install and setup of the Local Salesforce Architect MCP engine for GitHub Copilot, Cursor, and Claude Code. Fully local, offline Salesforce architecture tooling.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
One-click installer and configurator for the Local SF Architect engine — a
fully local, offline Model Context Protocol (MCP) server that gives your AI
assistant deterministic Salesforce architecture tools (governor-limit math,
Apex blast-radius analysis, architecture pattern search, diagrams, scoring, and
lint).
This extension does not replace the engine — it installs it from PyPI and
wires it into whichever AI agent you use:
Agent
How it is wired
GitHub Copilot / VS Code
Native MCP provider API (registerMcpServerDefinitionProvider, VS Code 1.101+); falls back to .vscode/mcp.json
Cursor
Merged into ~/.cursor/mcp.json (mcpServers)
Claude Code
claude mcp add --scope user (falls back to ~/.claude.json)
Requirements
VS Code (or Cursor) 1.101+
uv on your machine (the extension guides you if it is missing)
Internet access once for the engine install + the ~130 MB embedding model
What it does on first run
Detects your host (VS Code vs Cursor) and installed agents.
Installs the engine: uv tool install --upgrade sf-local-architect.
Downloads the embedding model and seeds the local knowledge base.
Wires the engine into every detected agent (existing configs are backed up).
After setup, open your AI chat in Agent mode and ask questions like:
"What breaks if I refactor AccountService.cls?"
"Will fetching 45,000 rows in a loop hit governor limits?"
"Score the architecture health of this project."
Commands
All under the Local SF Architect category in the Command Palette:
Run Setup (Install + Configure)
Install / Update Engine
Download Model + Seed Knowledge Base
Configure GitHub Copilot / VS Code · Cursor · Claude Code · All Detected Agents
sfArchitect.autoSetupOnActivate — run setup automatically on first activation
sfArchitect.autoConfigureAgents — wire detected agents without per-agent prompts
sfArchitect.autoDownloadModel — download the embedding model during setup
sfArchitect.uvPath — absolute path to uv (empty = auto-detect)
Privacy
The engine runs locally over stdio. Your code never leaves your machine. The
only network calls are the one-time PyPI install and embedding-model download.
Note: a marketplace icon.png (128×128+) should be added under media/ and
referenced from package.json before publishing.
