Compress OCR-heavy PDFs into dense packed images so agents can work with long visual documents.
{ "mcpServers": { "optical-context": { "command": "uvx", "args": [ "optical-context-mcp" ] } } }