Skip to content
| Marketplace
Sign in
Visual Studio Code>Machine Learning>rule42 FrameworkNew to Visual Studio Code? Get it now.
rule42 Framework

rule42 Framework

Rule42

|
372 installs
| (0) | Free
Agentic AI framework for the rule42 system: manage providers, models, memory, resilience, protocols, and toolkit settings.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

ANDL AI Configuration Extension

Professional AI configuration interface for ANDL with a polished, intuitive UI/UX. Manage AI providers, models, memory systems, resilience settings, and toolkit configuration through a modern webview interface.

Version: 0.3.0
Status: ✅ Production Ready

Configuration

The extension contributes a full settings schema under “ANDL AI” in VS Code Settings, organized to match the webview’s tabs:

  • AI & Models: default/cheaper model routing, execution mode
  • Chat & Context: prompt normalization, loop control, cognitive options
  • RAG & Indexing: STM/LTM/RAG, proactive context, response cache
  • Automation: edits/shell/git/loop/test toggles (schema only)
  • Security & Audit: redaction and audit logging
  • Budget & Costs: budget limits and token rates
  • Advanced: toolkit configuration and overrides
  • Orchestration: approvals and governance (schema only)
  • Integration: reporting, remote services, webhooks, reflector, attachments
  • System: sync, telemetry, cockpit URL, GitHub

Tip: You can edit settings either in the webview (Save/Apply) or directly in VS Code Settings. Changes persist across reloads.

Configure Providers (Pluggable)

This extension uses a centralized provider registry (src/extension/toolkit/providerRegistry.ts) as the single source of truth for all tool providers. No provider-specific logic is duplicated across consumer files.

Three-Tier Provider Discovery:

  1. Builtins — Extension-bundled providers (e.g., andl-indexer)
  2. Extensions — VS Code extensions via andl.ai.providerExtensions setting
  3. Modules — External modules via andl.ai.externalProviders setting or ANDL_AI_PROVIDERS env

For architecture details, see: src/extension/toolkit/README.md

Settings

  • andl.ai.providerExtensions (string[])
    • Provide extension IDs that export providers via one of:
      • getAiToolProviders(): Provider[]
      • aiToolProviders: Provider[]
      • default export: Provider[]
  • andl.ai.externalProviders (string[])
    • Provide Node module specifiers or file: URLs that export a provider through one of:
      • createToolProvider(): Provider
      • createOrchestratorProvider(...): Provider
      • default provider or named provider
  • Environment: ANDL_AI_PROVIDERS
    • Comma-separated module specifiers or file: URLs; same shapes as above.

When configured, the extension loads providers and enumerates tools via ai-client’s NodeToolkitAdapter.

Quick Add Integration (Marketplace-Style Discovery)

The Toolkit tab includes a ⚡ Quick Add Integration button that opens a marketplace-style modal showcasing pre-configured integrations:

Available Integrations (12 total):

  • Serper - Google Search (web, image, news, places) - $50 free credit
  • Resend - Email sending - 100/day free
  • Firecrawl - Web scraping - 500 pages/month free
  • Tavily - AI research - 1000 queries/month free
  • Browserless - Browser automation - 6 hours/month free
  • SerpAPI - Multi-engine search aggregator - $50 free credit
  • Brave Search - Privacy-focused search - $5/month for 2k queries
  • DuckDuckGo - Instant answers (free)
  • Perplexity - AI-powered research - $5/month starter
  • SendGrid - Transactional email - 100/day free
  • Postmark - Email delivery - 100/month free
  • Mailgun - Email API - 5k/month free

Coming Soon: 50+ additional integrations across 15+ categories

Features:

  • 🔍 Search integrations by name or description
  • 📂 Filter by category (Search, Communication, Data, Automation, Research)
  • 📊 View pricing tiers and free quotas
  • 🔑 Optional API key prompts (ESC to skip if not needed)
  • ⚡ One-click installation with automatic configuration

Tool Display:

  • Origin: Shows friendly service name (e.g., "Serper Google Search") instead of generic "external"
  • Risk Level: Each tool displays its risk rating (0-10 scale)

The extension bundles pre-configured provider templates that are automatically copied during build. No internet connection required for discovery - all metadata is local.

Contributing Integrations: Want to add your favorite API? See CONTRIBUTING_INTEGRATIONS.md for the complete guide.

Add from Module (Design-time Flow)

In the Toolkit tab, use "Add from Module…" to scan your workspace for module-based providers. The scanner looks for files named:

  • andl.tools.provider.js
  • andl.tools.provider.mjs
  • andl.tools.provider.cjs

Export shapes supported (any one):

  • export function createToolProvider(): Provider
  • export function createOrchestratorProvider(...): Provider
  • export const provider: Provider
  • export default Provider

Provider Pattern Support:

  • MCP-style: tools array with execute() methods
  • Legacy: listTools() async method with run() methods

After linking, the provider's tools are listed immediately. Use "Refresh Module Links" to re-scan linked modules for added/removed/updated tools.

Multi-Folder Workspace Support: The scanner loops through ALL workspace folders, not just the first. If automatic scan doesn't find your provider, use the "Browse for File…" button as a manual fallback.

Provider spec storage (workspace-relative by default)

When you link a module provider, the extension stores the spec as a workspace-relative value to keep settings portable across machines and directory layouts.

  • Stored format: workspace:<folderName>/<relative/path/to/file>
  • Examples:
    • workspace:andl-ai-client/examples/andl.tools.provider.mjs
    • Also supported at load time: ${workspaceFolder}/examples/andl.tools.provider.mjs and ${workspaceFolder:andl-ai-client}/examples/andl.tools.provider.mjs

At runtime, these specs are resolved to absolute file:// URIs for import. If the file is outside any open workspace folder, the absolute file:// URI is stored as a fallback.

Features

✨ Modern UI/UX - Material Design + Apple HIG principles with consistent 8pt grid system
🎨 Professional Design - Rounded cards, shadows, clear visual hierarchy
📊 Memory Stats - Live RAG index and LTM store statistics
📚 Comprehensive Help - Built-in guides for complex settings
🔧 Full Configuration - All ANDL AI settings in organized tabs
⚡ Real-time Updates - Instant feedback and validation

Quick Start

  1. Install the extension (.vsix file)
  2. Open Command Palette (Cmd+Shift+P / Ctrl+Shift+P)
  3. Run: "ANDL: Open AI Config"
  4. Configure your AI providers and settings
  5. Click Apply to save changes

Scripts

  • pnpm run compile — Production build (no source maps)
  • pnpm run package — Create .vsix file for installation
  • pnpm run dev — Development watch mode
  • pnpm test — Run tests

Profile & Protocol Architecture

Agent profiles and protocols use a packaged-builtin architecture with database seeding at runtime.

How it Works

  1. Canonical authoring: andl-ai-client owns the builtin profiles and protocol markdown corpus.
  2. Packaging: This extension bundles generated mirrors for VSIX/runtime seeding.
  3. Runtime: Profiles and protocols are seeded into the settings database on activation.

Canonical Sources

  • Profiles: ../andl-ai-client/src/profiles/builtins/ is the hand-edited canonical source.
  • Extension mirror: builtin-agent-profiles/ is generated from the canonical profile source and bundled into the VSIX.
  • Policies: ../andl-ai-client/andl/protocols/ is the hand-edited canonical source.
  • Extension policy mirror: andl/protocols/ in this extension is a generated mirror bundled into the VSIX.

The extension now derives reseed inventory from packaged mirrors rather than maintaining a second authoring source. ANDL no longer carries its own mirrored protocol tree.

Development Mode

For extension developers who want edits to persist to builtin sources:

Setting Description
andl.ai.development.syncToBuiltins Sync profile/protocol edits to builtin sources
andl.ai.development.forceReseed Force re-seed on next activation (auto-clears)

When syncToBuiltins is enabled, profile edits in Agent Studio write to BOTH the ai-client canonical source and the extension's packaged mirror. Policy edits write to the ai-client canonical source and the local packaged mirror.

Developer Workflow

The manual reseed loop stays the same, but profile syncing is now explicit and scriptable:

  • pnpm run sync:builtin-profiles — refresh builtin-agent-profiles/ from ../andl-ai-client/src/profiles/builtins/
  • pnpm run sync:builtin-policies — refresh packaged andl/protocols/ from ../andl-ai-client/andl/protocols/
  • pnpm run sync:builtins — refresh both profile and policy mirrors
  • pnpm run compile — now runs profile sync before building
  • ANDL: Reseed Agent Profiles / ANDL: Reseed Protocols & Policies — reseed workspaces from the packaged mirrors

Human-facing naming note: the project is moving toward calling these documents “policies”, but the runtime path remains andl/protocols/ and existing command IDs remain unchanged for compatibility.

Troubleshooting

Profiles/Protocols not found?

  1. Check for andl/profiles/ and andl/protocols/ under the configured ANDL Home
  2. Set andl.ai.development.forceReseed: true and reload VS Code
  3. Extension will re-seed from builtins

For full details, see: andl-ai-client/docs/PROFILE_ARCHITECTURE.md

Notes

  • This extension is a thin wrapper around an owned webview composition that imports UI components from src/ui/**.
  • The webview bundle (media/ai-config.js) is built locally with CSP-safe type="module"; inline source maps are enabled in dev only.
  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft