Web page
Interactive launch brief
Rendered in a sandboxed preview, not trapped as a file attachment.
Build anywhere. Dock to MiraDock.
The home for what your AI builds.
You used an AI coding tool to make something useful. MiraDock is where you put it: publish visual HTML Web pages and Markdown Bucket Docs, share them with people, and let agents read or write through MCP.
No GitHub repo, Vercel project, Supabase setup, or object storage bucket required for the artifact you just made.
Bucket workspace
One shareable object, visual first
Launch roomPUBLIC
Preview
Rendered HTML
Docs
Markdown
Sharing
Public/private
HTML
Markdown file
Implementation notes
Readable by people and retrievable by agents as native context.
Micro DB
Structured data layer
Part of the Bucket model, but not presented here as a live publishing action.
The Bucket
A workspace for the small things AI makes useful.
A Bucket is the core object: one shareable place for pages, docs, and context. It is meant to be opened visually by people and reached programmatically by agents.
1
Publish HTML and Markdown
MiraDock hosts Web pages as rendered HTML and Markdown files as Markdown, with images available for published work.
2
Share public or private Buckets
Buckets support public and private visibility, reusable share links, and public profile routes for discoverable work.
3
Browse what is public
The public Discover route lists public Buckets, featured demo Buckets, topic filters, creators, and HTML badges when a Bucket includes HTML pages.
Visual first
Lead with the rendered result.
MiraDock is not a code repository wearing a new label. HTML pages are meant to be inspected as pages. Markdown files are meant to be readable, linked, and reused as context.
Browse HTML BucketsHT
Web page
HTML
Ma
Markdown file
Markdown
MC
Agents
MCP
The wedge
The answer to “now where do I put it?”
1
Build anywhere
Use an AI coding tool to make a useful page, note, report, guide, or small interactive artifact.
2
Dock to MiraDock
Connect through MCP with a write token, then create a Bucket and push HTML or Markdown without setting up hosting infrastructure.
3
Share the Bucket
Open it in MiraDock, make it public or private, invite collaborators, or send a share link when the Bucket should stay controlled.
Collaboration
Share without losing the thread.
MiraDock ships collaboration around public Buckets through roles, forks, pull-request proposals, and audit-backed settings.
1
Collaborator roles
Public Buckets can invite collaborators as editors, reviewers, or readers.
2
Forks and attribution
Public Buckets can allow forks, show attribution, and list public forks.
3
Pull-request proposals
When enabled, signed-in users can propose changes to public Markdown files.
4
Audit trail
Public Bucket settings include audit history for important collaboration and visibility changes.
Agent-native
Buckets are made for people and agents.
Humans get a clean visual surface. Agents get MCP tools and AI-readable endpoints, using public access or tokens depending on the Bucket's visibility.
1
MCP write
Agents can create Buckets, push Markdown, push HTML, add images, and update metadata with the permissions you grant.
2
MCP read
Agents can read public Buckets and token-protected private Buckets as reusable context.
3
AI-readable routes
Public Buckets expose Markdown, /llm.txt, and /context.json surfaces for tools that fetch URLs.
Sharing modes
Public
People and agents can read public Buckets.
Private
Private Buckets stay controlled and require permission or tokens.
Share links
Create reusable links for controlled access without making a Bucket public.
Common questions
What is a Bucket?
A Bucket is a shareable workspace for the useful things an AI tool makes. Today that means rendered HTML Web pages, Markdown files, images, sharing, discovery, and MCP access.
Can I use Micro DB today?
No. Micro DB is part of the broader Bucket concept for structured data, but this homepage does not present it as a live feature.
Do I need GitHub, Vercel, Supabase, or S3?
No. MiraDock is for the moment after an AI tool made something useful and you need a durable place to put it.
How do agents connect?
Use the MCP setup guide to generate tokens and connect an MCP-aware tool. The homepage avoids hardcoding the endpoint because deployment URLs can change.
K
Give it a Bucket.
Save the useful thing your AI made in a durable workspace with sharing and agent access.
Start publishingPublish what your AI builds next.
Start with a Web page, a Markdown file, or a public Bucket worth sharing.