razorpay/slashdocs

What it solves

slashdocs provides a collaborative workspace designed specifically for AI agents and human teams to co-author technical documentation. It solves the problem of fragmented communication where agents generate reports or logs that are difficult for humans to review, comment on, or version-control effectively.

How it works

Agents generate self-contained HTML documents (containing text, diagrams, or dashboards) and push them to the slashdocs server via a CLI or MCP (Model Context Protocol) endpoint. These documents are stored as immutable versions. Humans can then view these documents in a web interface, leave anchored comments on specific sections, and resolve issues. Agents can read these comments to iterate on the document, creating a continuous feedback loop between automated systems and human reviewers.

Who it’s for

It is intended for engineering teams and organizations that integrate AI agents into their workflows for tasks like incident reporting, architecture reviews, data analysis, and automated testing. It is built for teams that want a structured, persistent home for agent-generated output.

Highlights

• Agent-First Design: Native support for MCP and a CLI to allow headless agents to publish and update documentation programmatically.
• Self-Contained Artifacts: Documents are stored as single HTML files, making them easy to version, archive, and render without external dependencies.
• Sandboxed Rendering: Uses separate hostnames for the application and document rendering to ensure security when displaying agent-generated content.
• Collaborative Workflow: Includes built-in commenting, version history, and Slack notifications to bridge the gap between AI output and human oversight.