# Markloop

> Markloop turns the HTML docs and artifacts you make with AI into one link your team or clients can comment on, without signing up. Your AI pulls the comments in and publishes the next version at the same link. 14-day free trial, no card required; reviewers are always free.

Markloop connects directly (over MCP) to Claude (chat, Cowork and Claude Code), ChatGPT (including Codex), Cursor and Grok; any other AI works through file upload and plain-file export. It works with single, self-contained HTML files. It is an independent product and is not affiliated with Anthropic or OpenAI.

## Product

- [Markloop homepage](https://markloop.io/): what it does, how the review loop works, pricing, FAQ
- [Agent setup](https://markloop.io/setup.md): step-by-step instructions for an AI agent to connect itself to Markloop over MCP, including sign-up for the 14-day free trial

## Docs

<!-- docs:start -->
- [Quickstart](https://markloop.io/docs/index.md): Connect your AI tool, publish a document, collect comments, and publish the next version in five steps.
- [Connect your AI tool](https://markloop.io/docs/connect.md): Add the Markloop MCP server to Claude, ChatGPT, Codex, Cursor, VS Code, Grok or another MCP client.
- [The review loop with your agent](https://markloop.io/docs/review-loop.md): How your AI agent pulls review comments, fixes the document, and publishes a new version at the same link.
- [Share a document](https://markloop.io/docs/sharing.md): Share one document with specific people, share a public link, or invite people to a project, and learn what each option lets reviewers do.
- [For reviewers](https://markloop.io/docs/for-reviewers.md): How to open a document someone shared with you in Markloop, leave comments and reply.
- [MCP tools reference](https://markloop.io/docs/mcp-tools.md): What your agent can do in Markloop: each tool, whether it reads or writes, its parameters and limits.
- [Changelog](https://markloop.io/docs/changelog.md): What changed in Markloop and its MCP server, newest first.
- [Data & security](https://markloop.io/docs/data-security.md): Who can see your documents, how previews are isolated, what a connected agent can do, and where your data is stored.
- [Troubleshooting & support](https://markloop.io/docs/troubleshooting.md): Fixes for common problems when an AI agent works with Markloop, and how to contact support.
<!-- docs:end -->

## Blog

- [How to Write a Client Proposal With AI and Get Comments (2026)](https://markloop.io/blog/write-client-proposal-with-ai/): a 5-step workflow for writing a client proposal with ChatGPT or Claude as one HTML page, checking it, and collecting the client's comments on one link
- [How to Write a Client Report With AI and Get Comments (2026)](https://markloop.io/blog/write-client-report-with-ai/): turn exports into a client report with ChatGPT or Claude, check the numbers, and use one comment link for a whole monthly engagement
- [HTML to Google Docs: How to Convert It and What Breaks (2026)](https://markloop.io/blog/html-to-google-docs/): a tested conversion of an AI-made HTML proposal to Google Docs - what survives (text, lists, tables, images), what breaks (layout, SVG charts, buttons), how to get HTML that converts cleanly, and when to share a link instead
- [Claude Artifact Comments: Who Can Comment and the Limits (2026)](https://markloop.io/blog/claude-artifact-comments/): who can comment on Claude and Claude Code artifacts, access levels, email-invite limits, why public links take no comments, and how to get comments from people without a Claude account
- [ChatGPT Sites Explained: Sharing, Limits and Feedback (2026)](https://markloop.io/blog/chatgpt-sites/): what ChatGPT Sites (formerly Codex Sites) is, who can open a Site, plans and limits, and how to get comments on a Site from people outside your workspace over MCP
- [How to Get Comments on a Cursor Canvas (2026)](https://markloop.io/blog/cursor-canvas-feedback/): Cursor shared canvases are team-only and read-only; how to turn a canvas into a standalone HTML file and get anchored feedback on it over MCP
- [Claude Artifacts Explained: How to Make One + 10 Examples](https://markloop.io/blog/what-are-claude-artifacts/): what Claude artifacts are, how to create one in the Claude app and Claude Code, 10 example prompts, limits, who can see them, and how they compare to ChatGPT Sites and Gemini Canvas
- [How to Share Claude Artifacts Outside Your Organization (2026)](https://markloop.io/blog/share-claude-code-artifacts/): step-by-step sharing from the Claude app and Claude Code, who can open the link on each plan, why public links take no comments, and 4 ways to get outside feedback
- [AI Agent Artifact Sharing Compared: Claude, Cursor, Codex (2026)](https://markloop.io/blog/claude-artifact-sharing-compared/): Claude artifacts, Cursor shared canvases, Codex Sites, hosting tools, and Markloop - compared by audience, feedback, and agent loop (updated October 2026)
- [Claude Live Artifacts Are Legacy: What Changed and What to Use Now (2026)](https://markloop.io/blog/claude-live-artifacts/): live artifacts became a legacy format on August 19, 2026 - who still gets them, how to edit old ones, and what to use now
- [HTML Is the New Markdown - and Reviewing It Is the Missing Half (2026)](https://markloop.io/blog/html-is-the-new-markdown/): why agents now write HTML instead of Markdown, and how the human review loop closes

## Legal

- [Terms of Service](https://markloop.io/terms-of-service)
- [Privacy Policy](https://markloop.io/privacy-policy)