Available Example Configurations

Available Example Configurations#

MassGen includes a comprehensive library of example configurations demonstrating various features, integrations, and use cases. Use them from any directory with the @examples/ prefix.

Quick Start#

# List all available configurations
massgen --list-examples

# Use any configuration from anywhere
massgen --config @examples/CATEGORY/SUBCATEGORY/FILENAME "Your question"

Categories#

Basic Examples#

Single Agent#

Simple configurations to get started with single AI agents.

@examples/basic/single/single_agent

  • Simple single agent configuration template for getting started

massgen --config @examples/basic/single/single_agent "What is machine learning?"

@examples/basic/single/single_gpt5nano

  • Single GPT-5-nano agent with reasoning, web search, and code execution enabled

massgen --config @examples/basic/single/single_gpt5nano "Calculate the first 100 prime numbers and plot their distribution"

@examples/basic/single/single_flash2.5

  • Single Gemini 2.5 Flash agent for quick tests and fast responses

massgen --config @examples/basic/single/single_flash2.5 "Explain quantum computing in simple terms"

@examples/basic/single/single_gemini2.5pro

  • Single Gemini 2.5 Pro agent for more complex reasoning tasks

massgen --config @examples/basic/single/single_gemini2.5pro "Analyze the economic impact of renewable energy adoption"

@examples/basic/single/single_gptoss120b

  • Single open-source GPT model (120B parameters) for local/privacy-focused use cases

Single Agent - Multimodal Capabilities#

@examples/basic/single/single_gpt4o_audio_generation

  • GPT-4o agent with audio output generation (text-to-speech)

massgen --config @examples/basic/single/single_gpt4o_audio_generation \
  "I want you to tell me a very short introduction about Sherlock Holmes in one sentence, and I want you to use emotion voice to read it out loud."

@examples/basic/single/single_gpt4o_video_generation

  • GPT-4o agent with video generation capabilities

massgen --config @examples/basic/single/single_gpt4o_video_generation \
  "Generate a 4 seconds video for 'Cherry blossom petals falling in the spring breeze, sunlight filtering through the pink petals creating a soft halo, slow motion capture, aesthetically beautiful and romantic, depth of field effect.'"

@examples/basic/single/single_gpt4o_image_generation

  • GPT-4o agent with image generation capabilities (DALL-E)

massgen --config @examples/basic/single/single_gpt4o_image_generation \
  "Generate an image of gray tabby cat hugging an otter with an orange scarf"

@examples/basic/single/single_gpt5nano_file_search

  • GPT-5-nano with file search/retrieval capabilities

massgen --config @examples/basic/single/single_gpt5nano_file_search \
  "What is humanity's last exam score for OpenAI Deep Research? Also, provide details about the other models mentioned in the PDF?"

@examples/basic/single/single_gpt5nano_image_understanding

  • GPT-5-nano with vision capabilities for image analysis

massgen --config @examples/basic/single/single_gpt5nano_image_understanding \
  "Please summarize the content in this image"

@examples/basic/single/single_openrouter_audio_understanding

  • Audio understanding via OpenRouter integration

@examples/basic/single/single_qwen_video_understanding

  • Qwen model for video analysis and understanding

Multi-Agent#

Multiple AI agents collaborating on tasks.

@examples/basic/multi/three_agents_default ⭐ Recommended

  • Three agents (Gemini 2.5 Flash, GPT-5-nano, Grok-3-mini) collaborating with web search enabled

massgen --config @examples/basic/multi/three_agents_default \
  "Analyze the pros and cons of renewable energy"

@examples/basic/multi/gemini_4o_claude

  • Gemini, GPT-4o, and Claude collaboration

massgen --config @examples/basic/multi/gemini_4o_claude \
  "What's best to do in Stockholm in October 2025"

@examples/basic/multi/gemini_gpt5nano_claude

  • Gemini, GPT-5-nano, and Claude collaboration

@examples/basic/multi/two_agents_gpt5

  • Two GPT-5 agents collaborating on tasks

@examples/basic/multi/two_agents_gemini

  • Two Gemini agents working together

@examples/basic/multi/geminicode_4o_claude

  • Gemini Code, GPT-4o, and Claude for coding tasks

@examples/basic/multi/geminicode_gpt5nano_claude

  • Gemini Code, GPT-5-nano, and Claude for coding tasks

@examples/basic/multi/glm_gemini_claude

  • GLM, Gemini, and Claude multi-model collaboration

@examples/basic/multi/gpt5nano_glm_qwen

  • GPT-5-nano, GLM, and Qwen working together

@examples/basic/multi/three_agents_opensource

  • Three open-source models working together

@examples/basic/multi/three_agents_vllm

  • Three agents using vLLM backend for high-performance inference

@examples/basic/multi/two_agents_opensource_lmstudio

  • Two open-source models via LM Studio

@examples/basic/multi/two_qwen_vllm_sglang

  • Two Qwen models using vLLM and SGLang backends

@examples/basic/multi/fast_timeout_example

  • Multi-agent setup with short timeout for quick demos

Multi-Agent - Multimodal#

@examples/basic/multi/gpt4o_audio_generation

  • Multiple agents with audio generation capabilities

@examples/basic/multi/gpt4o_image_generation

  • Multiple agents with image generation capabilities

@examples/basic/multi/gpt5nano_image_understanding

  • Multiple agents with vision capabilities

Provider-Specific Examples#

Azure#

@examples/providers/azure/azure_openai_single

  • Single agent using Azure OpenAI deployment

@examples/providers/azure/azure_openai_multi

  • Multiple agents using Azure OpenAI

OpenAI#

@examples/providers/openai/gpt5

  • GPT-5 model configuration

@examples/providers/openai/gpt5_nano

  • GPT-5-nano model configuration

Claude#

@examples/providers/claude/claude

  • Claude model configuration via Anthropic API

Gemini#

@examples/providers/gemini/gemini_gpt5nano

  • Gemini and GPT-5-nano hybrid configuration

Local#

@examples/providers/local/lmstudio

  • Using LM Studio for local model inference

massgen --config @examples/providers/local/lmstudio \
  "Explain machine learning concepts"

Other Providers#

@examples/providers/others/grok_single_agent

  • Single Grok agent via xAI API

@examples/providers/others/zai_coding_team

  • Team configuration using ZAI (Zhipu AI) models

@examples/providers/others/zai_glm45

  • GLM-4.5 model via ZAI

Tool Examples#

Code Execution Tools#

@examples/tools/code-execution/basic_command_execution

  • Basic command-line code execution with auto-detected virtual environments

massgen --config @examples/tools/code-execution/basic_command_execution \
  "Write a Python function to calculate factorial and test it"

@examples/tools/code-execution/multi_agent_playwright_automation

  • Multiple agents using Playwright for browser automation

massgen --config @examples/tools/code-execution/multi_agent_playwright_automation \
  "Browse three issues in https://github.com/Leezekun/MassGen and suggest documentation improvements. Include screenshots and suggestions in a website."

@examples/tools/code-execution/code_execution_use_case_simple

  • Simple use case demonstrating code execution workflow

@examples/tools/code-execution/docker_simple

  • Basic single-agent Docker execution (NEW in v0.0.32)

@examples/tools/code-execution/docker_multi_agent

  • Multi-agent Docker deployment with isolated containers (NEW in v0.0.32)

@examples/tools/code-execution/docker_with_resource_limits

  • Resource-constrained Docker setup with CPU/memory limits (NEW in v0.0.32)

@examples/tools/code-execution/docker_claude_code

  • Claude Code with Docker execution and automatic tool management (NEW in v0.0.32)

Filesystem Tools#

@examples/tools/filesystem/claude_code_single

  • Single Claude Code agent with full filesystem access

massgen --config @examples/tools/filesystem/claude_code_single \
  "Create a Python web scraper and save results to CSV"

@examples/tools/filesystem/claude_code_context_sharing

  • Demonstrates context sharing between agents with filesystem access

massgen --config @examples/tools/filesystem/claude_code_context_sharing \
  "Generate a comprehensive project report with charts and analysis"

@examples/tools/filesystem/gpt5mini_cc_fs_context_path

  • GPT-5-mini and Claude Code with context path configuration

massgen --config @examples/tools/filesystem/gpt5mini_cc_fs_context_path \
  "Enhance the website with: 1) A dark/light theme toggle with smooth transitions, 2) An interactive feature that helps users engage with the blog content, and 3) Visual polish with CSS animations"

@examples/tools/filesystem/gemini_gpt5nano_protected_paths

  • Demonstrates protected paths feature (read-only files within writable directories)

massgen --config @examples/tools/filesystem/gemini_gpt5nano_protected_paths \
  "Review the HTML and CSS files, then improve the styling"

@examples/tools/filesystem/gemini_gpt5nano_file_context_path

  • Gemini and GPT-5-nano with custom context paths

@examples/tools/filesystem/claude_code_flash2.5

  • Claude Code with Gemini 2.5 Flash filesystem tools

@examples/tools/filesystem/claude_code_gpt5nano

  • Claude Code with GPT-5-nano filesystem integration

@examples/tools/filesystem/cc_gpt5_gemini_filesystem

  • Claude Code, GPT-5, and Gemini with shared filesystem

@examples/tools/filesystem/grok4_gpt5_gemini_filesystem

  • Grok-4, GPT-5, and Gemini with filesystem tools

@examples/tools/filesystem/gemini_gemini_workspace_cleanup

  • Two Gemini agents for workspace management

@examples/tools/filesystem/gemini_gpt5_filesystem_casestudy

  • Case study of Gemini and GPT-5 filesystem collaboration

@examples/tools/filesystem/fs_permissions_test

  • Testing different filesystem permission levels

Filesystem Tools - Multi-turn#

@examples/tools/filesystem/multiturn/grok4_gpt5_claude_code_filesystem_multiturn

  • Multi-turn conversation with Grok-4, GPT-5, and Claude Code

@examples/tools/filesystem/multiturn/grok4_gpt5_gemini_filesystem_multiturn

  • Multi-turn conversation with Grok-4, GPT-5, and Gemini

@examples/tools/filesystem/multiturn/two_claude_code_filesystem_multiturn

  • Two Claude Code agents in multi-turn mode

@examples/tools/filesystem/multiturn/two_gemini_flash_filesystem_multiturn

  • Two Gemini Flash agents in multi-turn mode

MCP (Model Context Protocol) Integration#

@examples/tools/mcp/gpt5_nano_mcp_example

  • GPT-5-nano with MCP integration

massgen --config @examples/tools/mcp/gpt5_nano_mcp_example \
  "What's the weather forecast for New York this week?"

@examples/tools/mcp/multimcp_gemini

  • Gemini with multiple MCP servers (Requires BRAVE_API_KEY in .env)

massgen --config @examples/tools/mcp/multimcp_gemini \
  "Find the best restaurants in Paris and save the recommendations to a file"

@examples/tools/mcp/claude_mcp_example

  • Basic Claude with MCP integration

massgen --config @examples/tools/mcp/claude_mcp_example \
  "Research and compare weather in Beijing and Shanghai"

@examples/tools/mcp/gemini_mcp_example

  • Basic Gemini with MCP integration

@examples/tools/mcp/gemini_notion_mcp

  • Gemini with Notion MCP integration

@examples/tools/mcp/gemini_mcp_filesystem_test

  • Gemini with MCP filesystem tools

@examples/tools/mcp/gemini_mcp_filesystem_test_sharing

  • Gemini MCP filesystem with sharing between agents

@examples/tools/mcp/gemini_mcp_filesystem_test_single_agent

  • Single Gemini agent with MCP filesystem

@examples/tools/mcp/gemini_mcp_filesystem_test_with_claude_code

  • Gemini and Claude Code with shared MCP filesystem

@examples/tools/mcp/claude_code_simple_mcp

  • Simple MCP server integration with Claude Code

@examples/tools/mcp/claude_code_discord_mcp_example

  • Claude Code with Discord MCP server

@examples/tools/mcp/claude_code_twitter_mcp_example

  • Claude Code with Twitter MCP server

@examples/tools/mcp/gpt5mini_claude_code_discord_mcp_example

  • GPT-5-mini and Claude Code with Discord MCP

@examples/tools/mcp/grok3_mini_mcp_example

  • Grok-3-mini with MCP integration

@examples/tools/mcp/qwen_api_mcp_example

  • Qwen API with MCP integration

@examples/tools/mcp/qwen_local_mcp_example

  • Local Qwen with MCP integration

@examples/tools/mcp/gpt_oss_mcp_example

  • Open-source GPT with MCP integration

@examples/tools/mcp/five_agents_travel_mcp_test

  • Five agents with travel planning MCP tools

@examples/tools/mcp/five_agents_weather_mcp_test

  • Five agents with weather data MCP tools

Testing configs:

  • @examples/tools/mcp/claude_mcp_test

  • @examples/tools/mcp/gemini_mcp_test

  • @examples/tools/mcp/gpt5_nano_mcp_test

  • @examples/tools/mcp/grok3_mini_mcp_test

  • @examples/tools/mcp/qwen_api_mcp_test

  • @examples/tools/mcp/gpt_oss_mcp_test

Planning Mode#

@examples/tools/planning/five_agents_discord_mcp_planning_mode

  • Five agents with Discord MCP in planning mode

@examples/tools/planning/five_agents_filesystem_mcp_planning_mode

  • Five agents with filesystem MCP in planning mode

massgen --config @examples/tools/planning/five_agents_filesystem_mcp_planning_mode \
  "Create a comprehensive project structure with documentation"

@examples/tools/planning/five_agents_notion_mcp_planning_mode

  • Five agents with Notion MCP in planning mode

@examples/tools/planning/five_agents_twitter_mcp_planning_mode

  • Five agents with Twitter MCP in planning mode

@examples/tools/planning/gpt5_mini_case_study_mcp_planning_mode

  • Planning mode case study configuration

Team Configurations#

Pre-configured specialized teams for specific domains.

Creative Teams#

@examples/teams/creative/creative_team

  • Storyteller, Editor, and Critic agents optimized for creative writing and storytelling

massgen --config @examples/teams/creative/creative_team \
  "Write a short story about a robot who discovers music"

@examples/teams/creative/travel_planning

  • Specialized team for travel itinerary planning and recommendations

Research Teams#

@examples/teams/research/research_team

  • Information gatherer, domain expert, and synthesizer for research tasks

massgen --config @examples/teams/research/research_team \
  "Analyze market trends in renewable energy"

@examples/teams/research/news_analysis

  • Team specialized for analyzing news articles and current events

@examples/teams/research/technical_analysis

  • Team focused on technical documentation and code analysis

AG2 (AutoGen) Integration#

@examples/ag2/ag2_groupchat ⭐ Recommended

  • AG2 GroupChat with Coder, Reviewer, and Tester agents (entire group acts as one MassGen agent)

massgen --config @examples/ag2/ag2_groupchat \
  "Write a Python function to calculate factorial"

@examples/ag2/ag2_groupchat_gpt

  • AG2 GroupChat using GPT models

massgen --config @examples/ag2/ag2_groupchat_gpt \
  "Write a Python function to calculate factorial"

@examples/ag2/ag2_single_agent

  • Single AG2 agent demonstrating basic AG2 integration with MassGen

@examples/ag2/ag2_gemini

  • AG2 agent using Gemini backend

@examples/ag2/ag2_coder

  • AG2 coding agent with code execution capabilities

massgen --config @examples/ag2/ag2_coder \
  "Create a factorial function and calculate the factorial of 8. Show the result?"

@examples/ag2/ag2_coder_case_study

  • Case study of AG2 coding workflow

@examples/ag2/ag2_case_study

  • Comprehensive AG2 integration case study

Debug and Testing#

  • @examples/debug/skip_coordination_test - Skip coordination rounds for testing final presentation mode

  • @examples/debug/test_sdk_migration - Testing SDK migration features

  • @examples/debug/code_execution/command_filtering_blacklist - Code execution with command blacklist filtering

  • @examples/debug/code_execution/command_filtering_whitelist - Code execution with command whitelist filtering

  • @examples/debug/code_execution/docker_verification - Docker setup verification configuration

Path Format#

Use slashes (/) in @examples/ paths to match the actual directory structure:

# ✅ Correct - use slashes
massgen --config @examples/basic/single/single_gpt5nano "Question"
massgen --config @examples/tools/mcp/multimcp_gemini "Question"
massgen --config @examples/providers/claude/claude "Question"

# ❌ Incorrect - don't use underscores
massgen --config @examples/basic/single/single_gpt5nano_single_gpt5nano "Question"

Key Features Demonstrated#

Backend Integrations#

  • OpenAI (GPT-4o, GPT-5, GPT-5-nano, GPT-5-mini)

  • Anthropic Claude (Claude 3, Claude Code)

  • Google Gemini (2.5 Flash, 2.5 Pro, Gemini Code)

  • xAI Grok (Grok-3-mini, Grok-4)

  • Azure OpenAI

  • OpenRouter

  • Zhipu AI (GLM models)

  • Qwen (local and API)

  • Open-source models via LM Studio, vLLM, SGLang

Capabilities#

  • Reasoning (o1/o3-style extended thinking)

  • Web search integration

  • Code execution and interpretation

  • Filesystem operations (read/write/edit)

  • Audio generation and understanding

  • Image generation and understanding

  • Video understanding

  • Model Context Protocol (MCP) integrations

Advanced Features#

  • Multi-agent collaboration and orchestration

  • Context sharing between agents

  • Protected paths (read-only protection)

  • Custom context paths

  • Multi-turn conversations

  • AG2 (AutoGen) integration

  • GroupChat patterns

  • Workspace isolation and snapshot management

  • Docker execution mode with container isolation

Getting Started#

  1. Simple single agent: Start with @examples/basic/single/single_gpt5nano

  2. Multi-agent collaboration: Try @examples/basic/multi/three_agents_default

  3. Filesystem tools: Explore @examples/tools/filesystem/claude_code_single

  4. Specialized teams: Use @examples/teams/research/research_team for research tasks

  5. Advanced integration: Check out @examples/ag2/ag2_groupchat for AutoGen integration

Creating Custom Configurations#

All example configurations are located in massgen/configs/. You can:

  1. Copy an example that’s close to your needs

  2. Modify agent configurations, models, and capabilities

  3. Save to a custom location

  4. Use with massgen --config /path/to/your/config.yaml

For detailed configuration options, see the Configuration Guide.