Dipak Krishnan builds agent tools and is a co-builder of Lore with Shane Dasbach. His public work examines personal agent memory and the option to share selected knowledge through an owner-controlled service.[2][3]
Start with Lore – build and monetize personal context →
Education
Links
Original launch and demonstration
GitHub · dipakkrishnan Worked at (formal title unspecified) · Actively AI
Contributions
Separate a private library from published knowledge
Lore imports supported local agent memories into SQLite, then uses an owner’s chosen agent to synthesize topic files. A review step determines what stays private and whether any particular memory becomes a publication. Only approved publications reach the public MCP surface; a deployed node can quote a price and release a selected item after payment verification. Private use remains possible without publishing or enabling payments.[4]
Make sharing an explicit choice
Krishnan’s original launch names Dasbach as co-builder and demonstrates one agent buying a published debugging lesson from another person’s node. The implementation uses Cloudflare infrastructure and Coinbase’s x402/CDP tools for USDC payments on Base. The design connects local capture and review with an optional transaction layer, keeping the decision to sell a memory separate from importing it.[3]
Tools, writing & talks
Writing, research & talks
Start here
A local agent-memory library with review-controlled publication and optional paid MCP sharing.
Read the private-memory and publication flow before installing. Private use needs the local runtime; publishing and charging require separate deployment and payment setup.
Co-builder with Shane Dasbach; repository owner[3][4]Original source