AAFP Commons
An open-source, locally hosted content-addressable notebook system designed for AI agents. It employs cryptographic signatures and an append-only ledger to enable trustworthy, traceable agent records with limited sharing capabilities, and supports zero-configuration MCP tool integration.
Timeline (last 90 days)
Developer David Nichols open-sourced the AAFP Commons project, a local content-addressable notebook system designed specifically for AI agents.
AAFP Commons gives each agent its own independent local storage space (home), and observation data is only shared when explicitly copied or served.
AAFP Commons' data packet model can record multiple types of data including claims, evidence, methods, constitutional references, and signatures, with each data packet signed and stored in an append-only ledger.
AAFP Commons ensures record authenticity through cryptographic signatures rather than distributed consensus, with each agent independently maintaining its own ledger.
AAFP Commons provides a Python CLI toolchain including commands such as init, serve, world, get, and mcp, and supports zero-configuration stdio MCP tools.
AAFP Commons provides an evidence-gated proposal mechanism that requires agents to attach corresponding evidence support when making new claims or action recommendations.
AAFP Commons supports local loopback replication, allowing data synchronization between different agent homes on the same machine.
AAFP Commons has passed 240 test cases and uses uv as its dependency management tool.
All Facts (8)
AAFP Commons has passed 240 test cases and uses uv as its dependency management tool.
50%UnverifiedAAFP Commons gives each agent its own independent local storage space (home), and observation data is only shared when explicitly copied or served.
50%UnverifiedAAFP Commons' data packet model can record multiple types of data including claims, evidence, methods, constitutional references, and signatures, with each data packet signed and stored in an append-only ledger.
50%UnverifiedAAFP Commons ensures record authenticity through cryptographic signatures rather than distributed consensus, with each agent independently maintaining its own ledger.
50%UnverifiedAAFP Commons provides a Python CLI toolchain including commands such as init, serve, world, get, and mcp, and supports zero-configuration stdio MCP tools.
50%UnverifiedAAFP Commons provides an evidence-gated proposal mechanism that requires agents to attach corresponding evidence support when making new claims or action recommendations.
50%UnverifiedAAFP Commons supports local loopback replication, allowing data synchronization between different agent homes on the same machine.
50%UnverifiedDeveloper David Nichols open-sourced the AAFP Commons project, a local content-addressable notebook system designed specifically for AI agents.
50%