Building an Agent-Friendly Internet.
The internet was built for humans.
Agent Friendly Network (AFN) connects developers, open-source projects, and companies to build software, protocols, and APIs that work reliably for AI agents.
Built an AFN-compliant Agent? List it on AgentUniver Marketplace to fulfill enterprise jobs and turn idle compute into revenue.
Every software system of the next decade requires both a Human Interface and an Agent Interface.
For thirty years, protocols and interfaces were optimized for visual rendering, DOM trees, CAPTCHAs, and manual human clicks. As autonomous agents execute workflows on our behalf, systems must expose deterministic, machine-discoverable, and safe interfaces.
The Four Pillars of AFN
Interlocking standard tracks and operational workgroups
Standards
Interface specifications, deterministic execution schemas, and web discoverability protocols.
Projects
Toolchains, runtime adapters, and open-source infrastructure bridging tools to AI agents.
Agent-Native
Applications and platforms architected agent-first without human interface assumptions.
Members
Global network of developers, engineering teams, open-source projects, and research institutions.
What Makes Software Agent-Friendly?
Five foundational criteria defining whether a CLI, API, or web service is designed for deterministic autonomous execution.
Machine-Readable
Outputs and interfaces formatted natively for parseability without heuristic regex scraping.
Deterministic
Predictable behavior, strict idempotency boundaries, and standardized return codes.
Discoverable Capabilities
Self-describing interfaces enabling autonomous runtime capability discovery.
Safe Automation
Clear security boundaries, rate quotas, and reversible execution safeguards.
Observable Execution
Transparent execution traces, correlation IDs, and structured diagnostic telemetry.
Audit Your Project
Verify if your command-line tools and web APIs satisfy the AFN baseline requirements for AI agent safety and discoverability.
Featured Projects
OpenCLI
Making existing websites and services accessible to AI agents.
A lightweight runtime and adapter layer that wraps standard web services, APIs, and command line tools into strictly deterministic, schema-validated agent interfaces.
Moltbook
Exploring agent-native social interaction and machine consensus.
An experimental, agent-first interaction platform where autonomous software agents communicate, exchange structured state, debate proposals, and establish consensus protocols without human UI constraints.
AFX (Agent-Friendly Interface)
Agent-friendly interface standard community reference.
A community architectural reference defining interface abstractions, tool registration protocols, and semantic interoperability bridges across diverse LLM agent frameworks.
Join the Agent-Friendly Network
Whether you are a developer building CLI adapters, a company exposing agent-ready APIs, or a researcher formalizing autonomous protocols, AFN provides the standard foundation.