Getting Started
Installation
TODO: release on PyPI
Bypass installation with uvx
Check Available Coders
This shows which AI assistants are installed and available.
For example:
If you don't see the tick you will need to install the coding assistant.
Basic Usage
# Use default coder
metacoder "Your prompt here"
# Use specific coder
metacoder "Your prompt" --coder claude
# Specify working directory
metacoder "Analyze this project" --workdir ./myproject
# Use configuration file
metacoder "Build feature X" --config config.yaml
# Use custom instructions file
metacoder "Review code" --instructions guidelines.md
# Use MCP extensions
metacoder "Search for papers on LLMs" --mcp-collection research_mcps.yaml
Using MCP Extensions
Metacoder supports Model Context Protocol (MCP) servers that extend your AI coders with additional capabilities:
# Use all enabled MCPs from a collection
metacoder "Find recent AI papers" --mcp-collection mcps.yaml
# Enable specific MCPs only
metacoder "Search PMID:12345" --mcp-collection mcps.yaml --enable-mcp pubmed
# Combine with coder config
metacoder "Analyze database schema" --config claude_config.yaml --mcp-collection db_mcps.yaml
Next Steps
- Learn about supported coders
- Configure your assistants with configuration files
- Extend capabilities with MCP support