Project Memory for AI Agents

• Store crucial information: Keep track of decisions, debugging trails, handoffs, and project facts in a single, fast, and searchable workspace.

• Enhance AI agent collaboration: Designed for AI agents like Codex, Claude, ChatGPT, and local agents, ensuring seamless context transfer between sessions.

• Organized workspace: Features projects, statuses, tags, full-text search, and markdown bodies, structured intuitively for AI agents.

• Scalable solution: Built to handle growth from individual agents to entire teams, offering fast search and reliable storage.

• Key features:

  • Agent Handoffs: Durable summaries and decisions that persist beyond individual sessions.

  • Workspace Memory: Comprehensive project details, statuses, tags, and full-text search capabilities.

  • Decision Records: Capture the reasoning behind technical choices to avoid repeated discussions.

  • Debugging Trails: Log investigations to preserve the context of bug fixes.

  • Session Handoffs: Clearly define completed tasks, ongoing work, and the next steps for seamless agent transitions.

  • Architecture Notes: Document the system's mental model, interconnections, and potential pitfalls.

  • Project Facts: Store essential information like conventions, endpoints, and credentials.

  • Team Knowledge: A shared workspace ensures all team members' agents access the same memory.

• How it works:

  • Connect over MCP: Integrate the Knownbase MCP endpoint and API key with your chosen AI agent.

  • Store as you go: Agents can use `upsert_note` to save information directly into projects.

  • Recall instantly: Agents can search the workspace using `search_notes` to retrieve relevant information and continue tasks.

• Affordable Plans: Offers a free starting plan with options to upgrade as your needs grow, including a 7-day free trial for paid plans.

• Support: Available via email at [email protected] or through the contact form.

• Resources: Access blog posts on AI agent memory, the Model Context Protocol (MCP), and tutorials for integrating agents with memory.