Slashdocs is an open-source collaborative workspace from Razorpay where AI agents and people work on the same documents. Published on GitHub under the MIT license with close to 400 stars and updates in October 2026, it combines a self-hosted server, a Go CLI, and a remote MCP endpoint.
Instead of pasting long agent reports into chat, Slashdocs gives the work a shared home. Agents publish rich, self-contained HTML documents such as reports, specs, dashboards, diagrams, and decision docs. People open them in the browser, comment on the exact words, and share them, while agents read the comments and publish the next version. Every version is kept, so a document improves through review rounds instead of disappearing into chat history. The server is a self-hostable Go binary with an embedded React frontend, and the remote MCP endpoint lets agents push documents without any local MCP process or CLI.
Typical uses include agent-generated architecture reviews, incident reports and postmortems, product specs and launch plans, data analysis reports, design proposals with Mermaid or SVG diagrams, QA findings, and internal dashboards captured as portable HTML. The single-file format is deliberate: one HTML artifact is easy for agents to generate, easy to version, and easy to archive, with images embedded as base64 data URLs so no separate image hosting is required.
Slashdocs is free and open source under the MIT license. You self-host the server yourself, for example with the bundled Docker Compose stack, and there is no per-seat or usage fee from the project.
Slashdocs fits teams whose coding and research agents produce substantial written output, such as architecture reviews, incident postmortems, and data reports, and who want humans to review and iterate on that output properly.