Skilldocs

Product Introduction

1. Definition: Skilldocs is a collaborative, real-time markdown editor and documentation platform designed for technical teams. It functions as a specialized SaaS (Software-as-a-Service) tool that integrates live multi-user editing, inline commenting, and a direct feedback loop with AI coding agents via the Model Context Protocol (MCP).
2. Core Value Proposition: Skilldocs exists to streamline the technical review and collaboration process for software development teams. Its primary value is closing the loop between human code review and AI-assisted development by providing a shared, version-controlled environment where markdown-based skills, documentation, and code snippets can be edited, discussed, and then seamlessly fed back to an AI agent like Claude Code for implementation.

Main Features

1. Real-Time Collaborative Markdown Editing: Skilldocs provides a live, multi-cursor editing environment where multiple team members can edit the same markdown document simultaneously. How it works: It uses operational transformation (OT) or a similar conflict-free replicated data type (CRDT) technology to synchronize text changes in real-time across all connected clients, displaying each user's cursor and edits live.
2. Inline Commenting & Threaded Discussions: The platform allows users to leave comments anchored to specific quotes or lines within the markdown. These comments form threaded conversations, creating a contextual feedback log directly tied to the document's content, which is crucial for code and documentation review.
3. MCP Server Integration for AI Agent Loop: This is the core technical integration. Skilldocs operates as an MCP (Model Context Protocol) server. How it works: Developers can connect Claude Code (or other MCP-compatible agents) to Skilldocs using a workspace-specific API token. The agent can then use tools like skilldocs_open to push a local file, skilldocs_read to get the document and comment threads, and skilldocs_pull to write the updated document back, creating a fully automated review-and-apply cycle.
4. GitHub Repository Synchronization: Teams can connect Skilldocs directly to a GitHub repository (including private repos). The platform scans the repo for markdown files (skills), making them instantly browsable and editable within the workspace. It supports updating a baseline version directly from GitHub, ensuring the reviewed document aligns with the shipped code.
5. Version Control & Diff Viewing: Skilldocs includes built-in document versioning. Users can mark versions as baselines, view a full version history, and see a precise diff of all changes made since the last marked version. This feature is essential for tracking the evolution of a skill or document through collaborative edits.

Problems Solved

1. Pain Point: Disconnected and inefficient feedback loops for technical documentation and AI-generated code (skills). Traditional methods involve sharing markdown files via chat, email, or PR comments, leading to fragmented conversations and manual copy-pasting of feedback for AI agents.
2. Target Audience: The primary user personas are Software Engineering Teams, Developer Advocates creating technical content, AI Engineers building and refining prompts and skills for LLMs, and Technical Product Managers who need to review and approve specification documents. It is specifically built for organizations using Google Workspace for authentication.
3. Use Cases:
   • Collaborative AI Skill Development: A team refining a complex prompt for an AI agent can edit the markdown skill together in real-time, discuss edge cases in comments, and then have the AI agent directly read the finalized version and all feedback.
   • Technical Documentation Review: Engineering teams can collaboratively write and review internal API documentation, architecture decision records (ADRs), or runbooks with live cursors and contextual comments.
   • Code-Agent Workflow Integration: A developer using Claude Code locally can instantly push a generated skill to Skilldocs for team review, collect asynchronous feedback via comments, and then command the agent to read the diff and comments to apply the changes automatically.

Unique Advantages

1. Differentiation: Unlike generic collaborative editors (Google Docs, Notion) or code-centric platforms (GitHub PRs), Skilldocs is purpose-built for the intersection of markdown-based technical content and AI-assisted development. It combines the real-time ease of a Google Doc with the versioning and repository integration of a developer tool, and uniquely adds a bidirectional bridge for AI agents.
2. Key Innovation: The proprietary Skilldocs MCP Server is the key innovation. It transforms the platform from a passive document store into an active participant in the AI development workflow. By providing structured tools (skilldocs_read, skilldocs_reply, skilldocs_pull), it enables AI agents to not just receive a static document but to engage dynamically with the entire collaboration history and context.

Frequently Asked Questions (FAQ)

1. What is Skilldocs used for? Skilldocs is used for real-time collaborative editing of markdown documents, specifically designed for teams developing and reviewing AI agent skills, technical documentation, and code-related content, with direct integration to feed changes and feedback back to AI coding assistants.
2. How does Skilldocs integrate with Claude Code? Skilldocs integrates with Claude Code through a custom MCP (Model Context Protocol) server. After setup, Claude Code can use specific Skilldocs tools to push documents, read collaborative edits and comments, reply to feedback threads, and pull the finalized text back to the local environment.
3. Is Skilldocs secure for private company data? Yes, Skilldocs supports private GitHub repositories and requires Google Workspace authentication for company workspaces, ensuring access is limited to users with a verified company email domain. API tokens for MCP access are scoped per workspace and stored server-side.
4. Can I use Skilldocs without GitHub? Yes, you can use Skilldocs without GitHub by creating new documents directly within the platform or by uploading .md files. The GitHub integration is an optional feature for syncing with existing repository-based markdown skills.
5. What is the "agent loop" in Skilldocs? The "agent loop" refers to the continuous workflow where a developer pushes a local skill to Skilldocs, the team collaborates on it, and then the AI agent (like Claude Code) reads the resulting diff and comments to understand the required changes, applies them, and can potentially push a new version, closing the collaboration loop between humans and AI.