# sf-intelligence: offline Salesforce MCP server

> sf-intelligence: offline Salesforce MCP server is a free MCP server for sales and crm. Offline Salesforce impact analysis from your own metadata, answered with cited components. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Free
- **Platforms:** CLI, self-hosted
- **Website:** https://sfi.auditforce.cloud
- **Repository:** https://github.com/pranavnagrecha/salesforce-intelligence
- **MCP Registry name:** io.github.PranavNagrecha/salesforce-intelligence
- **Categories:** sales and CRM, developer tools
- **Listing updated:** 2026-11-15

## About

Instead of connecting to a live org, sf-intelligence answers from a local vault, an `org-kb` folder built from retrieved metadata. You point the `SFI_VAULT` environment variable at that folder; without it the server falls back to `./org-kb` in whatever directory the host launched in, and reports what it looked for through `sfi.setup_status`.

Answers are structured around change risk. A field-deletion check returns a verdict such as safe, risky, blocking or review, lists the Apex, formulas and roll-ups that depend on the field, and spells out which metadata types were found empty and which were not checked. Save-order questions list triggers, validation rules, flows and approvals in the sequence they run. A typo-tolerant resolver asks a clarifying question when a name is ambiguous.

By default the assistant sees 16 tools: eight answer tools and eight navigation tools, with other analyses behind the `sfi.run_analysis` gateway. A built-in demo org, a fictional solar installer, lets you try it without a Salesforce login. The optional live read-only plane stays off until enabled. Its site reports 8 of 8 correct answers on admin questions against a real org, a measurement the vendor made itself.

## Features

- **Field deletion verdicts:** Returns safe, risky, blocking or review with the components behind each reason.
- **Save-order explanation:** Lists triggers, validation rules, flows and approvals in the order they run for a record save.
- **Effective permissions:** Works out who can see or edit an object from profiles and permission sets.
- **Dead Apex detection:** Finds classes and triggers nothing calls, tiered as definitely, likely or uncertain.
- **Coverage disclosure:** Each answer says which metadata kinds were checked and which were not in the snapshot.
- **Typo-tolerant resolver:** Asks which component you meant when a name matches several.
- **Built-in demo org:** Try the server on a sample org without logging in to Salesforce.

## Good for

- Check what depends on a custom field before deleting it
- Trace what runs when a record is saved before changing automation
- Answer who can edit an object during a permissions review
- Find unused Apex classes to clean up
- Explore an unfamiliar org through an AI assistant

## Install

### Claude Code (Terminal)

```sh
claude mcp add salesforce-intelligence --env SFI_VAULT='<SFI_VAULT>' -- npx -y sf-intelligence
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "salesforce-intelligence": {
      "command": "npx",
      "args": [
        "-y",
        "sf-intelligence"
      ],
      "env": {
        "SFI_VAULT": "<SFI_VAULT>"
      }
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "salesforce-intelligence": {
      "command": "npx",
      "args": [
        "-y",
        "sf-intelligence"
      ],
      "env": {
        "SFI_VAULT": "<SFI_VAULT>"
      }
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "salesforce-intelligence": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "sf-intelligence"
      ],
      "env": {
        "SFI_VAULT": "<SFI_VAULT>"
      }
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "salesforce-intelligence": {
      "command": "npx",
      "args": [
        "-y",
        "sf-intelligence"
      ],
      "env": {
        "SFI_VAULT": "<SFI_VAULT>"
      }
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "salesforce-intelligence": {
      "command": "npx",
      "args": [
        "-y",
        "sf-intelligence"
      ],
      "env": {
        "SFI_VAULT": "<SFI_VAULT>"
      }
    }
  }
}
```

## Tools

- `sfi.resolve`: Resolves messy component names to the right Salesforce component.
- `sfi.search_components`: Searches components in the vault.
- `sfi.get_component`: Fetches details of one component.
- `sfi.health_check`: Checks the server and vault health.
- `sfi.list_analyses`: Lists the analyses available through the gateway.
- `sfi.describe_analysis`: Describes one analysis.
- `sfi.run_analysis`: Gateway tool that runs any analysis not exposed directly.
- `sfi.live_consent`: Controls consent for the optional live read-only plane.
- `sfi.setup_status`: Reports where the server looked for the vault and what it found.

## FAQ

### Does it change my Salesforce org?

No. The site and README say it never writes to your org and runs on your computer.

### What do I need to set?

Set SFI_VAULT to the absolute path of your org-kb vault; the server starts without it but has no org to answer about.

### Which assistants does it work with?

Claude Code, Claude Desktop, Cursor, VS Code with Copilot and Codex.

### Can I try it without an org?

Yes, `npx -y sf-intelligence demo` starts a built-in demo org with no Salesforce login.

### Is it free?

The site describes it as free; no paid plan is listed.

### Is sf-intelligence: offline Salesforce MCP server free?

Yes. sf-intelligence: offline Salesforce MCP server is free to use.

### How do I add sf-intelligence: offline Salesforce MCP server to Claude Code?

Run this command in your terminal: claude mcp add salesforce-intelligence --env SFI_VAULT='<SFI_VAULT>' -- npx -y sf-intelligence

### What are the best alternatives to sf-intelligence: offline Salesforce MCP server?

The closest alternatives to sf-intelligence: offline Salesforce MCP server are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.

## Alternatives

- [REA](https://ai-powered.com/mcp/rea): REA is an open-source MCP server for developer tools.
- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [GitHub](https://ai-powered.com/mcp/github-2): GitHub is an open-source MCP server for developer tools.
- [Serena MCP: the IDE for your agent](https://ai-powered.com/mcp/serena-mcp-the-ide-for-your-agent): Serena MCP: the IDE for your agent is an open-source MCP server for developer tools.
- [XERJ](https://ai-powered.com/mcp/xerj): XERJ is an open-source MCP server for knowledge and memory.
- [Ouroboros](https://ai-powered.com/mcp/ouroboros): Ouroboros is an open-source MCP server for developer tools.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [Concord MCP](https://ai-powered.com/mcp/concord-mcp): Concord MCP is a freemium MCP server for developer tools.

---
Source: https://ai-powered.com/mcp/sf-intelligence-offline-salesforce-mcp-server
