Outsource MCP

1# Outsource MCP
2 
3An MCP (Model Context Protocol) server that enables AI applications to outsource tasks to various model providers through a unified interface.
4 
5<img width="1154" alt="image" src="https://github.com/user-attachments/assets/cd364a7c-eae5-4c58-bc1f-fdeea6cb8434" />
6 
7<img width="1103" alt="image" src="https://github.com/user-attachments/assets/55924981-83e9-4811-9f51-b049595b7505" />
8 
9 
10Compatible with any AI tool that supports the Model Context Protocol, including Claude Desktop, Cline, and other MCP-enabled applications.
11Built with [FastMCP](https://github.com/mcp/fastmcp) for the MCP server implementation and [Agno](https://github.com/agno-agi/agno) for AI agent capabilities.
12 
13## Features
14 
15- 🤖 **Multi-Provider Support**: Access 20+ AI providers through a single interface
16- 📝 **Text Generation**: Generate text using models from OpenAI, Anthropic, Google, and more
17- 🎨 **Image Generation**: Create images using DALL-E 3 and DALL-E 2
18- 🔧 **Simple API**: Consistent interface with just three parameters: provider, model, and prompt
19- 🔑 **Flexible Authentication**: Only configure API keys for the providers you use
20 
21## Configuration
22 
23Add the following configuration to your MCP client. Consult your MCP client's documentation for specific configuration details.
24 
25```json
26{
27  "mcpServers": {
28    "outsource-mcp": {
29      "command": "uvx",
30      "args": ["--from", "git+https://github.com/gwbischof/outsource-mcp.git", "outsource-mcp"],
31      "env": {
32        "OPENAI_API_KEY": "your-openai-key",
33        "ANTHROPIC_API_KEY": "your-anthropic-key",
34        "GOOGLE_API_KEY": "your-google-key",
35        "GROQ_API_KEY": "your-groq-key",
36        "DEEPSEEK_API_KEY": "your-deepseek-key",
37        "XAI_API_KEY": "your-xai-key",
38        "PERPLEXITY_API_KEY": "your-perplexity-key",
39        "COHERE_API_KEY": "your-cohere-key",
40        "FIREWORKS_API_KEY": "your-fireworks-key",
41        "HUGGINGFACE_API_KEY": "your-huggingface-key",
42        "MISTRAL_API_KEY": "your-mistral-key",
43        "NVIDIA_API_KEY": "your-nvidia-key",
44        "OLLAMA_HOST": "http://localhost:11434",
45        "OPENROUTER_API_KEY": "your-openrouter-key",
46        "TOGETHER_API_KEY": "your-together-key",
47        "CEREBRAS_API_KEY": "your-cerebras-key",
48        "DEEPINFRA_API_KEY": "your-deepinfra-key",
49        "SAMBANOVA_API_KEY": "your-sambanova-key"
50      }
51    }
52  }
53}
54```
55 
56Note: The environment variables are optional. Only include the API keys for the providers you want to use.
57 
58## Quick Start
59 
60Once installed and configured, you can use the tools in your MCP client:
61 
621. **Generate text**: Use the `outsource_text` tool with provider "openai", model "gpt-4o-mini", and prompt "Write a haiku about coding"
632. **Generate images**: Use the `outsource_image` tool with provider "openai", model "dall-e-3", and prompt "A futuristic city skyline at sunset"
64 
65## Tools
66 
67### outsource_text
68Creates an Agno agent with a specified provider and model to generate text responses. 
69 
70**Arguments:**
71- `provider`: The provider name (e.g., "openai", "anthropic", "google", "groq", etc.)
72- `model`: The model name (e.g., "gpt-4o", "claude-3-5-sonnet-20241022", "gemini-2.0-flash-exp")
73- `prompt`: The text prompt to send to the model
74 
75### outsource_image
76Generates images using AI models.
77 
78**Arguments:**
79- `provider`: The provider name (currently only "openai" is supported)
80- `model`: The model name ("dall-e-3" or "dall-e-2")
81- `prompt`: The image generation prompt
82 
83Returns the URL of the generated image.
84 
85> **Note**: Image generation is currently only supported by OpenAI models (DALL-E 2 and DALL-E 3). Other providers only support text generation.
86 
87## Supported Providers
88 
89The following providers are supported. Use the provider name (in parentheses) as the `provider` argument:
90 
91### Core Providers
92- **OpenAI** (`openai`) - GPT-4, GPT-3.5, DALL-E, etc. | [Models](https://platform.openai.com/docs/models)
93- **Anthropic** (`anthropic`) - Claude 3.5, Claude 3, etc. | [Models](https://docs.anthropic.com/en/docs/about-claude/models/overview)
94- **Google** (`google`) - Gemini Pro, Gemini Flash, etc. | [Models](https://ai.google.dev/models)
95- **Groq** (`groq`) - Llama 3, Mixtral, etc. | [Models](https://console.groq.com/docs/models)
96- **DeepSeek** (`deepseek`) - DeepSeek Chat & Coder | [Models](https://api-docs.deepseek.com/api/list-models)
97- **xAI** (`xai`) - Grok models | [Models](https://docs.x.ai/docs/models)
98- **Perplexity** (`perplexity`) - Sonar models | [Models](https://docs.perplexity.ai/guides/model-cards)
99 
100### Additional Providers
101- **Cohere** (`cohere`) - Command models | [Models](https://docs.cohere.com/v2/docs/models)
102- **Mistral AI** (`mistral`) - Mistral Large, Medium, Small | [Models](https://docs.mistral.ai/getting-started/models/models_overview/)
103- **NVIDIA** (`nvidia`) - Various models | [Models](https://build.nvidia.com/models)
104- **HuggingFace** (`huggingface`) - Open source models | [Models](https://huggingface.co/models)
105- **Ollama** (`ollama`) - Local models | [Models](https://ollama.com/library)
106- **Fireworks AI** (`fireworks`) - Fast inference | [Models](https://fireworks.ai/models?view=list)
107- **OpenRouter** (`openrouter`) - Multi-provider access | [Models](https://openrouter.ai/docs/overview/models)
108- **Together AI** (`together`) - Open source models | [Models](https://docs.together.ai/docs/serverless-models)
109- **Cerebras** (`cerebras`) - Fast inference | [Models](https://cerebras.ai/models)
110- **DeepInfra** (`deepinfra`) - Optimized models | [Models](https://deepinfra.com/docs/models)
111- **SambaNova** (`sambanova`) - Enterprise models | [Models](https://docs.sambanova.ai/cloud/docs/get-started/supported-models)
112 
113### Enterprise Providers
114- **AWS Bedrock** (`aws` or `bedrock`) - AWS-hosted models | [Models](https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html)
115- **Azure AI** (`azure`) - Azure-hosted models | [Models](https://learn.microsoft.com/en-us/azure/ai-foundry/concepts/foundry-models-overview)
116- **IBM WatsonX** (`ibm` or `watsonx`) - IBM models | [Models](https://www.ibm.com/docs/en/software-hub/5.1.x?topic=install-foundation-models)
117- **LiteLLM** (`litellm`) - Universal interface | [Models](https://docs.litellm.ai/docs/providers)
118- **Vercel v0** (`vercel` or `v0`) - Vercel AI | [Models](https://sdk.vercel.ai/docs/introduction)
119- **Meta Llama** (`meta`) - Direct Meta access | [Models](https://www.llama.com/get-started/)
120 
121### Environment Variables
122 
123Each provider requires its corresponding API key:
124 
125| Provider | Environment Variable | Example |
126|----------|---------------------|---------|
127| OpenAI | `OPENAI_API_KEY` | sk-... |
128| Anthropic | `ANTHROPIC_API_KEY` | sk-ant-... |
129| Google | `GOOGLE_API_KEY` | AIza... |
130| Groq | `GROQ_API_KEY` | gsk_... |
131| DeepSeek | `DEEPSEEK_API_KEY` | sk-... |
132| xAI | `XAI_API_KEY` | xai-... |
133| Perplexity | `PERPLEXITY_API_KEY` | pplx-... |
134| Cohere | `COHERE_API_KEY` | ... |
135| Fireworks | `FIREWORKS_API_KEY` | ... |
136| HuggingFace | `HUGGINGFACE_API_KEY` | hf_... |
137| Mistral | `MISTRAL_API_KEY` | ... |
138| NVIDIA | `NVIDIA_API_KEY` | nvapi-... |
139| Ollama | `OLLAMA_HOST` | http://localhost:11434 |
140| OpenRouter | `OPENROUTER_API_KEY` | ... |
141| Together | `TOGETHER_API_KEY` | ... |
142| Cerebras | `CEREBRAS_API_KEY` | ... |
143| DeepInfra | `DEEPINFRA_API_KEY` | ... |
144| SambaNova | `SAMBANOVA_API_KEY` | ... |
145| AWS Bedrock | AWS credentials | Via AWS CLI/SDK |
146| Azure AI | Azure credentials | Via Azure CLI/SDK |
147| IBM WatsonX | `IBM_WATSONX_API_KEY` | ... |
148| Meta Llama | `LLAMA_API_KEY` | ... |
149 
150**Note**: Only configure the API keys for providers you plan to use.
151 
152## Examples
153 
154### Text Generation
155```
156# Using OpenAI
157provider: openai
158model: gpt-4o-mini
159prompt: Write a haiku about coding
160 
161# Using Anthropic
162provider: anthropic
163model: claude-3-5-sonnet-20241022
164prompt: Explain quantum computing in simple terms
165 
166# Using Google
167provider: google
168model: gemini-2.0-flash-exp
169prompt: Create a recipe for chocolate chip cookies
170```
171 
172### Image Generation
173```
174# Using DALL-E 3
175provider: openai
176model: dall-e-3
177prompt: A serene Japanese garden with cherry blossoms
178 
179# Using DALL-E 2
180provider: openai
181model: dall-e-2
182prompt: A futuristic cityscape at sunset
183```
184 
185## Development
186 
187### Prerequisites
188 
189- Python 3.11 or higher
190- [uv](https://github.com/astral-sh/uv) package manager
191 
192### Setup
193 
194```bash
195git clone https://github.com/gwbischof/outsource-mcp.git
196cd outsource-mcp
197uv sync
198```
199 
200### Testing with MCP Inspector
201 
202The MCP Inspector allows you to test the server interactively:
203 
204```bash
205mcp dev server.py
206```
207 
208### Running Tests
209 
210The test suite includes integration tests that verify both text and image generation:
211 
212```bash
213# Run all tests
214uv run pytest
215```
216 
217**Note:** Integration tests require API keys to be set in your environment.
218 
219## Troubleshooting
220 
221### Common Issues
222 
2231. **"Error: Unknown provider"**
224   - Check that you're using a supported provider name from the list above
225   - Provider names are case-insensitive
226 
2272. **"Error: OpenAI API error"** 
228   - Verify your API key is correctly set in the environment variables
229   - Check that your API key has access to the requested model
230   - Ensure you have sufficient credits/quota
231 
2323. **"Error: No image was generated"**
233   - This can happen if the image generation request fails
234   - Try a simpler prompt or different model (dall-e-2 vs dall-e-3)
235 
2364. **Environment variables not working**
237   - Make sure to restart your MCP client after updating the configuration
238   - Verify the configuration file location for your specific MCP client
239   - Check that the environment variables are properly formatted in the configuration
240 
241## Contributing
242 
243Contributions are welcome! Please feel free to submit a Pull Request.
244