Liminal

Liminal provides a collaborative workspace designed for seamless interaction between humans, AI agents, and teams. It functions as an AI-native second brain, allowing users to bring their own agents and maintain ownership of their files. The core purpose is to streamline workflows that involve AI agents by offering a unified and accessible platform for knowledge management and collaboration. The problem Liminal addresses stems from the evolving nature of work where AI agents are becoming primary interfaces. Traditional tools like Notion and Google Docs are not optimized for this shift, often leading to slow and expensive interactions with AI models. The maker experienced friction when their agent, Claude Code, worked best with local markdown files, a format that existing collaboration tools struggled to handle efficiently, thus necessitating a new approach to workspace management. One key feature is the ability to work directly with your agent, not just through it. When an agent writes files to disk in markdown or HTML, Liminal instantly renders them as a clean user interface. This allows for natural review and editing within a browser, eliminating the struggle of wrestling with raw markdown code and providing a more intuitive user experience. Liminal also facilitates instant sharing without friction. A lightweight CLI syncs the workspace to the cloud in real time. By sharing a link, teammates can access the same live workspace, ensuring that edits made by one person or agent are instantly reflected on another's disk. This eliminates the need for cumbersome processes like sending markdown files over Slack, hosting HTML, or relying on slow cloud storage solutions like Google Drive. The product acts as a shared second brain for every agent on the team. Since files reside locally, each agent has direct access to all the information the team has written, decided, and learned. This direct access ensures no external calls or lag, preserving context between sessions and making knowledge readily available for AI agents. Furthermore, Liminal operates without a middleman, avoiding what the maker calls 'MCP tax.' Direct file access means no intermediate model calls, no proprietary formats, and no wasted tokens. This approach makes context cheap, fast, and consistently fresh, optimizing the efficiency of AI-driven workflows. Liminal's overall approach is to bridge the gap between local file formats that AI agents understand and the need for real-time, collaborative editing and viewing. It leverages a lightweight CLI for synchronization and a browser-based interface for rendering and editing, creating a cohesive environment for human-agent collaboration. The benefits for users include a more efficient workflow when collaborating with AI agents, reduced costs associated with AI model interactions, and the creation of a continuously updated, shared knowledge base. By owning their files and having direct agent access, users can build a more robust and accessible 'second brain' for their team. Specific use cases include directing an AI agent to write files in markdown or HTML, which Liminal then renders for immediate review and editing. Teams can collaborate on documents, share project updates, or collectively build a knowledge base, with all changes syncing in real-time. This is particularly useful for workflows where AI agents are central to content creation and management. Liminal is designed for individuals and teams who are heavily invested in using AI agents as their primary interface for work. It is particularly beneficial for those who find traditional note-taking and document collaboration tools inadequate for agent-based workflows. The product is free to use, accessible via web, and utilizes technologies that enable real-time synchronization and local file management. In essence, Liminal redefines collaborative workspaces by centering them around AI agents and local file formats, creating a dynamic and efficient shared second brain for modern teams.