MCP read

Read public Buckets from AI tools

Public MiraDock Buckets expose stable AI-readable endpoints for agents that fetch URLs or connect through MCP-aware workflows.

Open a public Bucket

Use Discover or Topics to find a public Bucket you want an AI tool to read.

Browse discover

Use the AI endpoints

Each public Bucket provides /llm.txt for a compact guide, /context.json for structured context, and markdown rendering for human-readable Markdown files. HTML Web pages appear with text snippets and content_type metadata.

Choose native or text reads

MCP get_web_page defaults to native content. Pass format text when an agent needs extracted text from HTML Web pages instead of sanitized HTML.

Read Micro DB tables

MCP DB tools can list tables, read schema, query typed rows with validated filters and sorting, and fetch a row by id when the Bucket is readable.

Create your own source

To publish a Bucket that other agents can read, connect an AI tool with a write token and push Bucket parts through MCP.

Set up MCP write