A schema-enforced module standard for the AI-Perceivable era — so agents can act, not guess. Your schema becomes the contract — every surface, one definition, zero glue.
Agents burn most of their reasoning guessing how to call APIs — a cognitive tax on every action. apcore removes it by design.
Modules, interfaces, and tools an agent reads directly — no guessing parameters, no parsing prose docs. A clean flow from perception → cognition → execution that measurably improves agent efficiency.
Licensed under Apache 2.0. Fully open, transparent, and community-driven.
Language-agnostic specifications built on high-performance bridges like PyO3 and Zod. Built in Python, TypeScript, and Rust today.
Strict input/output contracts on every interface. Predictable, validatable, and self-documenting.
Built for real-world use with AI guidance for self-healing agents, monitoring, and enterprise-grade observability.
Adapters for MCP, gRPC, HTTP, WebSocket, and more. Integrate with any protocol or framework.
The AI-perceivable module standard — specification, protocol bridges, and developer toolkit.
Currents, reefs, and migrations — a living ecosystem built on the apcore standard.
The apcore family is a reef of species — each mascot maps to a role in the AI-perceivable stack.
One schema, one reef. Every species speaks the same contract.
Start with apcore to adopt the AI-Perceivable module standard, use apcore-mcp to bridge them to AI agents, or jump into apflow for task orchestration. All open source and production ready.
