@sf-ai-toolkit/core

Core engine for SF AI Toolkit.

This package contains all the logic used by the CLI and VS Code extension — scanning, file generation, inline diagnostics, MCP bootstrap, drift detection, and more. Use it if you want to build your own tooling on top of SF AI Toolkit.

Installation

npm install @sf-ai-toolkit/core

Usage

import {
  scanProject,
  planSetup,
  applySetup,
  analyseFile,
  detectFileType,
  bootstrapMcp,
  detectDrift,
} from '@sf-ai-toolkit/core';

// Scan a Salesforce DX project and get an AI readiness score
const result = await scanProject('/path/to/sf/project');
console.log(result.score);        // 0–100
console.log(result.missing);      // files and config that are missing
console.log(result.recommendations);

// Plan and apply the setup
const plan = await planSetup('/path/to/sf/project', { preset: 'core' });
const applied = await applySetup('/path/to/sf/project', plan);
console.log(applied.filesCreated);
console.log(applied.filesSkipped); // existing files are never overwritten

// Run inline diagnostics on an Apex file
const fileType = detectFileType('MyClass.cls'); // 'apex'
const diagnostics = analyseFile(apexSourceCode, fileType);
// diagnostics: [{ ruleId, message, severity, line, ruleFile }]

// Bootstrap MCP config for Cursor and Claude Code
const mcp = await bootstrapMcp('/path/to/sf/project', { orgAlias: 'my-sandbox' });
// writes .cursor/mcp.json and .mcp.json

What Is Included

| Module | Description | |--------|-------------| | scanProject | Scans a project and returns an AI readiness score across 21 signals | | planSetup | Plans which files to create for a given preset, without writing anything | | applySetup | Applies a plan — creates missing files, never overwrites existing ones | | analyseFile | Runs inline diagnostic rules on Apex, LWC JS, and LWC HTML source | | detectFileType | Classifies a file path as apex, lwc-js, lwc-html, or unknown | | getHoverContent | Returns explanation and fix suggestion for a given diagnostic rule ID | | bootstrapMcp | Generates .cursor/mcp.json and .mcp.json for a Salesforce org | | validateMcpConfig | Validates an MCP config file for common mistakes | | detectDrift | Checks whether AI setup files still contain the expected content signals | | checkTeamSync | Compares local files against a remote team config | | fetchTeamConfig | Fetches a team config JSON from a URL | | readOrgContext | Reads the current default org alias from project config files | | detectAgentforceContext | Scans force-app/ for Agentforce metadata | | buildDeployPreview | Classifies components in force-app/ and identifies deployment risks | | listInstalledSkills | Returns installed Cursor skills from .cursor/skills/ | | generateClaudeMemModeJson | Generates a claude-mem Salesforce DX mode JSON file | | generateReadinessReport | Formats a scan result as a readable text report |

Presets

| Preset | Description | |--------|-------------| | core | Standard Salesforce DX project | | lwc | Adds extra LWC rules and skills | | agentforce | Adds Agentforce and AFV Library support | | data-cloud | Adds Data Cloud docs and rules | | experience-cloud | Adds Experience Cloud rules |

Requirements

• Node.js 18 or later

Related

• VS Code Extension — install this if you want the editor experience
• @sf-ai-toolkit/cli — install this if you want terminal commands

Author

Nikhil Karkra

License

MIT