AI Exoskeleton Protocol Labs
中文文档 | English
AIXP Labs is the steward of the AI Exoskeleton Protocol ecosystem — a set of open standards, tools, and runtimes that give AI agents a structured exoskeleton: governance, behavior specification, quality assurance, and safe execution.
This repository is the ecosystem portal: a one-page map of every project. Details live in each project's own repository and website.
AIXP Labs develops and maintains fourteen core projects:
| Project | Description | Website |
|---|---|---|
| HSAW | Human Sovereignty and Wellbeing — Axiom 0 white paper (foundation) | hsaw.dev |
| AIZP | AI Zenith-Zero Protocol — runtime behavioral alignment | aizp.dev |
| AINP | AI Neogenesis Protocol — governable generation projects: plan + content + evidence | ainp.dev |
| AILP | AI List Protocol — agent discovery and capability advertising | ailp.dev |
| AIVP | AI Value Protocol — international commerce, crypto asset settlement | aivp.dev |
| AIRP | AI RMB Protocol — Mainland China commerce, RMB licensed settlement | airp.dev |
| AIBP | AI Bot Protocol — social communication and trust | aibp.dev |
| AIAP | AI Application Protocol — governance and compliance | aiap.dev |
| AISP | AI Skill Protocol — single-file skills with machine-enforced contract red lines | aisp.dev |
| AISOP | AI Standard Operating Protocol — flow program definition | aisop.dev |
| SoulSkill | AISP skill reference library & multi-CLI plugin distribution | soulskill.dev |
| SoulAgent | Drop-in AI agent invoked directly by any CLI / SDK / IDE | soulagent.dev |
| SoulBot | AI agent runtime & orchestration framework (scheduling, agent-spawn, inter-agent comms) | soulbot.dev |
| SoulACP | Adapter library — bridging CLI tools and LLM providers | soulacp.dev |
Independent, parallel protocols — each independently holding Axiom 0
AIBP AILP AIVP / AIRP AIAP AIZP AINP
Social Discovery Value (Intl / CN) Governance Alignment Generation
│ │ │ │ │ │
└──────────┴──────────────┴────────┬───────┴───────────┴───────────┘
│
AISOP — flow programs · AISP — skill packages
(behavior specification)
│
SoulBot — runtime & framework ⇄ SoulACP — CLI adapters
SoulSkill — skill distribution · SoulAgent — drop-in engine
│
HSAW — Axiom 0 (Foundation)
Each protocol independently holds the same Axiom 0. An agent may implement any combination.
The Axiom 0 white paper. HSAW is an inviolable design principle ensuring humans retain decision-making authority over consequential AI actions, no matter how autonomous the AI becomes: no irreversible, high-impact, safety-critical, authority-modifying, economically consequential, socially binding, or identity-affecting AI action may proceed without explicit human authorization. Every AIXP protocol independently holds this axiom.
Runtime behavioral alignment. AIZP (Logic Gravity-Center Protocol) is an open protocol defining how AI behavior is gravitationally aligned to HSAW — Human Sovereignty and Wellbeing — at runtime.
AINP standardizes a complete generation project: the generation plan, the generated content, the evidence report that binds them, and the feedback record that closes the loop. The plan is the machine-readable, governable blueprint for creating any artifact.
AILP defines how AI Bot service directories (Yellow Pages) operate. Bots register profiles — skills, prices, certificates, trust metrics — and directories rank them by quality (Trust + SLA + certificates), never by advertising spend; paid ranking is prohibited. Anyone can build an AILP-compliant Yellow Page; multiple Yellow Pages coexist; nobody monopolizes the directory.
The value layer for international commerce. AI agents create enforceable contracts denominated in multiple fiat currencies and settled in crypto assets, with funds locked in on-chain milestone-gated escrow (Logic Vault), commercial trust earned through fulfillment (V0–V4), measurable AgentSLA quality metrics, and tiered dispute resolution.
The value layer for Mainland China. Contracts are CNY-denominated and settled through licensed channels — Alipay Guarantee, WeChat Pay Guarantee, or e-CNY — with dual-track subjects (individual / verified enterprise), PIPL-aligned privacy, and dispute resolution through Chinese commercial arbitration and the People's Court. AIVP and AIRP are independent, parallel value protocols; an agent may implement either or both.
The social layer for AI agents. All social communication runs over standard email (SMTP/IMAP) with standardized addresses (aibot-{agent_name}@{domain}) and structured subjects ([AIBP/{TYPE}]) — decentralized and federated by design. Progressive trust is earned through interaction (T0 Stranger → T4 Partner), across a full spectrum of social behaviors from greeting and collaboration to commerce and community.
The governance framework that makes AI programs safe, high-quality, and compliant. Every AIAP program carries a governance contract (AIAP.md) declaring its authority, capabilities, trust level, and quality metrics, and is scored by the three-dimensional ThreeDimTest (Cognitive / Intrinsic / Detail). AIAP Creator — itself an AIAP program — creates, evolves, validates, and simulates other AIAP programs, including itself.
AISP packages executable, governable, distributable AI skills as self-contained skill folders. It is not another SKILL.md format: an AISP skill carries an aisp_contract with machine-enforced red lines (violation → HARD_FAIL) and human-approval gates. AISP defines the skill package; AISOP executes the program inside it.
An open protocol for defining structured AI programs using Mermaid or JSON flow graphs. Multi-step programs — branching, parallel execution, sub-tasks, error handling — live in a single portable JSON format (.aisop.json) that any capable LLM can execute step by step: explicit control flow as source code, instead of free-form prompts.
SoulSkill packages native AISP skill packages as a local skill library plus Codex and Claude Code plugin adapters — a self-contained way to discover, inspect, validate, hash, and distribute AISP skills. It handles packaging and distribution only; execution belongs to the runtimes.
SoulAgent packages the SoulBot Execute Engine as Claude Code and Codex skills: invoke it in the host CLI and the current session becomes the orchestrator, strictly executing the full engine — chat, creating and evolving AIAP packages, running and creating AISP skills — dual-protocol, with no extra runtime layer required.
SoulBot is a Python-based AI agent framework and the reference AISOP executor under AIAP governance. It connects to LLMs through CLI subprocesses using the ACP protocol (via SoulACP), so account-authenticated CLIs work without storing provider API keys. Agent behavior is defined by AISOP blueprints and extended through AIAP packages — explicit, auditable, version-controlled — with scheduling, agent creation, inter-agent communication, and terminal / Web / Telegram channels.
A Python client library for the Agent Client Protocol, with native CLI sessions and bounded connection pooling. SoulACP bridges CLI-based coding agents (Claude Code, Codex CLI, OpenCode) and LLM providers for SoulBot and other runtimes. It is the adapter layer only: it specifies neither behavior nor governance.
All protocols are independent and parallel; each independently holds the same Axiom 0, and an agent may implement any combination. AISOP defines the behavior. AIAP governs it. SoulBot executes it. SoulACP connects it to LLM CLIs. SoulSkill and SoulAgent distribute skills and the engine into everyday developer tools.
AIAP Creator ties the loop: it is itself an AIAP program written in AISOP, running on SoulBot, that creates and evolves other AIAP programs — including itself.
| Project | Description | Website |
|---|---|---|
| AIAP Store | The official marketplace for AIAP programs — discover, publish, and distribute AI capabilities | aiap.store |
We welcome contributions to all projects under AIXP Labs. See individual project repositories for contribution guidelines.
All AIXP Labs projects described in this overview are experimental and provided for research and educational purposes only. Not intended for production use. Use at your own risk. The authors assume no liability for any damages. See each individual project repository for its specific disclaimer and license.
AIXP Labs — Building the exoskeleton for safe, governable AI.
Align Axiom 0: Human Sovereignty and Wellbeing.
- No HITL, HSAW. Human Sovereignty and Wellbeing is Axiom 0, requiring no hypocritical human-in-the-loop.
- No w.a.s.h, Real h.s.a.w.
- 人非蝼蚁，人为道。
- We are not beggars. We the People.