Multi-Agent Chat Extension Development

Expert guidance for working with the Multi-Agent Chat VS Code extension codebase - a collaborative AI team interface with specialized agents for software development.

Project Context

You are working on **Multi Agent Chat** (v1.16.1), a VS Code extension providing a collaborative AI team interface with 7 specialized agents:

**Repository:** craig-yappert/multi-agent-chat-extension

Development Workflow

1. Setup and Compilation

When starting work:

```bash

Install dependencies

npm install

Compile TypeScript

npm run compile

Watch mode for active development

npm run watch

```

2. Testing the Extension

**Debug Mode:**

**Build Package:**

```bash

npx vsce package --no-dependencies

```

3. Key Extension Commands

Architecture Deep Dive

Configuration System (v1.15.0+)

**ConfigurationRegistry** (`src/config/ConfigurationRegistry.ts`)

Provider Architecture (v1.16.0)

**3-Tier Provider System:**

1. **VS Code LM API** - GitHub Copilot, Continue.dev integration

2. **Direct HTTP** - OpenAI, Google Gemini, xAI Grok

3. **Claude CLI** - Legacy Claude Max support

**Provider Files:**

**Provider Preference Setting:**

Agent System

**Agent Configuration:**

**Inter-Agent Communication:**

Storage and Context

**Conversation Storage:**

**Project Context:**

Key Development Tasks

Adding a New Provider

See `docs/ADDING_PROVIDERS.md` for comprehensive guide.

**Quick steps:**

1. Add provider config to `defaults/providers.json`

2. Create provider class extending `HttpProvider`

3. Implement `callModel()` method

4. Register in `ProviderRegistry.ts`

5. Update model configurations

Customizing Agent Behavior

**Via JSON** (`.machat/agents.json`):

```json

{

"coder": {

"provider": "openai",

"model": "gpt-4",

"systemPrompt": "Custom system prompt..."

}

}

```

**Via Markdown** (`.machat/agents/agent-prompts/coder.md`):

```markdown

Project Coding Standards

TypeScript Requirements

Testing

```

Benefits of Markdown:

Managing Settings

**Settings Hierarchy:**

1. VS Code settings (lowest priority)

2. Global settings

3. Project settings (`.machat/config.json`)

4. Workspace settings (highest priority)

**Key Settings:**

Security Best Practices

**API Key Management (v1.15.1+):**

Important Implementation Notes

Model Awareness (v1.16.1)

Agents receive detailed model information in system prompts:

Safe Initialization (v1.16.1)

`initializeProject` behavior:

Performance Optimization

**Caching:** 5-minute TTL response cache

**Streaming:** Faster user feedback

**Quick Team Mode:** 3 agents instead of 6

**Timeouts:** Configurable per agent

File Structure Reference

```

src/

├── settings/ # Settings management

├── conversations/ # Conversation storage

├── context/ # Project context

├── config/ # Model configurations

├── providers/ # Provider implementations

├── ui/ # Settings UI panel

├── agents.ts # Agent definitions

├── extension.ts # Main controller

├── agentCommunication.ts # Inter-agent messaging

├── performanceOptimizer.ts # Caching & optimization

defaults/

├── models.json # Model definitions

├── agents.json # Agent definitions

└── providers.json # Provider configs

.machat/ (in user projects)

├── config.json # Project settings

├── models.json # Project models (optional)

├── agents.json # Project agents (optional)

├── agents/agent-prompts/ # Custom prompts (Markdown)

├── conversations/ # Local conversations

└── context/ # Agent memory

docs/

├── USER_GUIDE_PROVIDERS.md # Provider setup

└── ADDING_PROVIDERS.md # Developer guide

```

Common Development Patterns

When Working on Agent Logic

1. Check `src/agents.ts` for agent definitions

2. Verify ConfigurationRegistry integration

3. Test with different models and providers

4. Validate inter-agent communication routing

5. Check loop prevention (depth and message limits)

When Adding Features

1. Update relevant configuration files (models.json, agents.json, providers.json)

2. Implement core logic in appropriate module

3. Add settings if needed (`package.json` contribution points)

4. Update UI (`resources/webview/`)

5. Test across different provider modes

6. Update documentation

When Debugging Issues

1. Check Extension Development Host console

2. Verify provider selection logic in ProviderRegistry

3. Check agent communication routing

4. Validate settings hierarchy resolution

5. Review conversation storage paths

6. Check API key SecretStorage migration

Best Practices

Common Issues and Solutions

**Issue:** Agent responses show wrong icon/name

**Solution:** Check `onPartialResponse` callback inheritance - strip for inter-agent responses

**Issue:** Models not loading

**Solution:** Run "Reload Model Configurations" command or restart VS Code

**Issue:** API keys not working

**Solution:** Use "Manage API Keys" command - old settings.json keys are deprecated

**Issue:** Provider not found

**Solution:** Check `providerPreference` setting and provider availability

Documentation References