ai-standards 开源项目推荐（26 个）

hiero-ledger /hiero-consensus-specifications

Canonical specifications for Hiero Consensus Specifications (HCS) — originally written and maintained by Hashgraph Online

AthenaCore /AwesomeResponsibleAI

一份精心策划的优秀学术研究、书籍、伦理守则、课程、数据库、数据集、框架、机构、成熟度模型、通讯、原则、播客、法规、报告、负责任规模政策、工具和标准的列表，内容涵盖负责任、可信和以人为本的人工智能。

synaptiai /prompt-decorators

A standardized framework for enhancing how LLMs process and respond to prompts through composable decorators, featuring an official open standard specification and Python reference implementation. Claude Code plugin and MCP server integration.

ai-domain-data /spec

The open, vendor-neutral format for authoritative domain data consumed by AI systems, search, and other automated agents

agent-manifest /agent-manifest

Declarative specification for AI agents to publish identity, capabilities and operational boundaries before interaction.

ali-ulu /awesome-agent

Awesome Agent SDLC v1.2 — Open-source standard documentation repository for AI-agent-assisted software development: agent roles, the A2A handoff protocol, the JUDGE verification layer, a maturity model, and multi-AI-tool (Cursor, Gemini, Claude, Copilot) configurations. Licensed AGPL-3.0.

Hollow-house-institute /Hollow_House_Standards_Library

Canonical AI governance standards library and terminology registry for Behavioral AI Governance and Execution-Time Governance systems.

giorgiokito /WAI

Web Agent Interface, a domain standard json file to expose public information and endpoint to AI Agents

zenprocess /standra

standra.ai — Open standards and tooling for AI-assisted engineering

zenprocess /cacp

Compressed Agent Communication Protocol — structured I/O format for LLM agents (~200 vs ~2000 tokens)

padosoft /ai-standards-kit

Enterprise AI Engineering Standards & Agents toolkit for multi-stack development

AlbertShamin /controlled-ai-core

Independent AI governance and control standard

lostlight530 /china-agentic-observatory

Bilingual, source-grounded observatory of China's AI and agent ecosystem across history, policy, standards, research, open source, infrastructure, industry, applications, evaluation, safety, and governance.

zenprocess /afterburn

Extract residual intelligence from spent Claude Code sessions

ftessier /cred

A memorable standard for Human-AI attribution

attahwonder /faias-standard

A baseline for reviewing generative, agentic and predictive AI systems in financial institutions: 109 review questions with test procedures and source registers. CC BY 4.0.

OpenSynapseLabs /Open-Synapse-AI-Standard-OSAS-

Principled standard for building trustworthy AI systems through architecture, constitutional RL and rigorous verification.

zenprocess /servingcard

该项目在 GitHub 上没有填写描述

metronisys /metronisys-certification

Metronisys Certification for commercial use

zenprocess /pawbench

PawBench - 4-dimensional LLM inference benchmark. Multi-turn, multi-agent, parallel dispatch with tool calling. Inspired by Lola (@_justlolathings).