Markloop is a web application designed to facilitate the review of HTML documents created by AI agents. Its primary function is to serve as a feedback loop, allowing human reviewers to comment on and answer questions within the context of a document. The tool then consolidates this feedback and sends it back to the AI agent in a structured format, enabling the agent to apply revisions before the work proceeds. Markloop is specifically intended for reviewing HTML documents such as specifications, reports, and proposals, and is not designed for live websites or shipped applications. The application supports a workflow that begins with document ingestion. Users can upload HTML files directly or allow their AI agent to send documents via the Model Context Protocol (MCP). Once a document is in Markloop, it is stored as a versioned file within a project. Users can invite viewers to a project for collaborative commenting or share a single read-only file link for someone who only needs to review. Reviewers can pin comments to specific sections or sentences, and these comments remain anchored to the exact version of the document on which they were made. The agent can then pull all feedback back over MCP as structured context, including the target, quote, intent, and version, eliminating the need for manual copy-pasting. Key features include unlimited projects, documents, and versions for a single creator seat, with unlimited free reviewers. The tool offers native MCP integration for Claude Code and Codex, allowing agents to push and pull feedback without leaving the terminal. Other agents can also access the same feedback as plain files. For team use, a plan provides up to five creator seats, a shared workspace, team roles and permissions, the ability to remove Markloop branding, and priority support. The tool is available without requiring a credit card and can be canceled at any time. Typical use cases involve scenarios where coding agents draft documents that require human understanding, review, or approval. The tool addresses common problems such as formatting loss when documents are moved to other platforms, scattered feedback across emails and messages, and the manual effort of translating comments back into prompts for an AI agent. By keeping feedback anchored to the document and structured for the agent, Markloop aims to streamline the revision process for agent-made HTML documents.