Hey PH! 👋 I'm Aditya, a backend engineer building agentic AI systems at a financial firm. Every time you start a new agent session, you're back to square one. Re-explaining context, re-setting rules,redefiningg scope. It's frustrating and repetitive, and it doesn't scale. I built ContextBook to fix that. It's not memory (pre-loaded, ambient, noisy). It's context — surgical, on-demand, scoped exactly to what the agent needs at that moment. Think of it as a structured knowledge layer your agents can pull from, not a dump they swim through. The best part? It's completely platform-agnostic. One ContextBook instance works across Claude, GPT, Gemini, or any MCP-compatible agent — on any platform, any workflow. Build once, use everywhere. How it works: Organize knowledge into Books (scoped containers) Break it into Pages (semantic chunks ≤1000 words) Let agents retrieve exactly what's relevant via 8 MCP tools Powered by Voyage AI embeddings + pgvector for fast semantic search It's open-source, MCP-native, and deploys in minutes. Stop re-explaining yourself to every agent. Give them the right context, right when they need it — across every platform you build on. Would love your feedback, questions, and brutal honesty. 🙏

ContextBook

Stop re-explaining. Give agents the right context

Stop re-explaining. Give agents the right context