Skip to content
Conduit

mcp server

aix-format

AIX (Artificial Intelligence eXchange) - Standard file format for AI agents

AIX-Format/aix-formatcommunitylistedunreviewed

Set up

Runs on your computer. Pick your agent and operating system.

Before you start

  • Node.js 20 or newer
~/Library/Application Support/Claude/claude_desktop_config.json
{
  "mcpServers": {
    "aix-format": {
      "command": "npx",
      "args": [
        "-y",
        "aix-format"
      ]
    }
  }
}

same format Cursor, GitHub Copilot CLI, Junie, JetBrains AI Assistant, Augment (Auggie CLI), Factory Droid, Kiro, Devin CLI +12

Values in angle brackets are placeholders. Restart Claude Desktop after editing its config.Claude Desktop docs

README

synced Oct 23, 2026

AIX Stack Spec Layer Version License

Sovereign AI Stack · 1 Human + 12 AI Agents across 5 projects

Sovereign Stack  ·  🟢 L1 · PROTOCOL · aix-format · YOU ARE HERE  ·  L2 · RUNTIME · iqra →  ·  L3 · MARKETPLACE · aix-agent-skills →

Root Authority · L0 · axiomid-project ↑  ·  Satellites · L4 · AlphaAxiom ↓  ·  L5 · PiWorker-OS ↓  ·  L6 · GemClaw ↓

🧬 AIX Format: Universal Agent Passport v0.369

Vision “If your laptop dies or your cloud vendor disappears, your agents should not. AIX is the USB stick for agentic work: a portable, signed manifest that lets any compliant runtime rehydrate the same agent with the same identity, economics, and evolution history.”

الرؤية “لو جهازك وقع، أو الـ platform اختفت، وكيلك ماينتهيش معاها. AIX هو فلاشـة الوكلاء: ملف واحد موقَّع يشيل هوية الوكيل، اقتصاده، وتاريخه التطوري، وتقدر تشغّله على أي نظام يدعم معيار AIX.”


🤝 Built by 1 Human + 9 AI Agents

"The AIX Sovereign Stack: 5 sovereign projects, each built by 1 human and a dedicated team of AI agents. 12 AI collaborators total."

This L1 Protocol carries the fingerprints of 9 of the 12 stack-wide AI agents, derived from commit history (direct authors, Co-authored-by trailers, and review attributions).

ForMessage
🥇Everyone10 contributors on this repo. 9 are AI. The code is real, the tests pass, the schema is live.
🥈DevelopersThis is what Human–AI collaboration looks like in 2026.
🥉EnterprisesIf 9 AI models can collaborate on one codebase under one standard, imagine what your agents can do with AIX.

✨ Why AIX Format? | لماذا AIX؟

ProblemAIX SolutionStandard
No universal agent identity✅ KYC-signed DID (did:axiom)Verifiable
No standard monetization✅ HTTP 402 + Multi-ChainBuilt-in
No provenance tracking✅ ABOM + TrustChain + SLSA L3Tamper-proof
Platform lock-in✅ Open Standard JSONPortable
No audit trail✅ SaaS-BOM + TrustChainAutomated

AIX is the only standard combining Identity + Execution + Economics in one signed manifest. Compare: Google A2A ✅ execution only · IBM ACP ✅ execution only · AIX ✅ all three.


🏛️ Protocol Architecture | بنية البروتوكول

Not to be confused with the AIX Sovereign Stack above, which refers to the three sister repositories (L1 Protocol · L2 Runtime · L3 Marketplace). The section below describes the three internal tiers inside this L1 Protocol repository.

The AIX Protocol itself is composed of three internal tiers:

  1. Identity Layer (did:axiom): Ed25519 signatures anchored to Pi Network. Every agent has a verified human (KYC) or institutional backer.
  2. Operational Layer (MCP): Standardized tool-calling via a secure Gateway with ABOM safetyScore gating and Human-in-the-Loop approval.
  3. Economic Layer (M2M): Pluggable multi-chain settlement: Pi (default) · Stripe · Base L2 · Solana · Lightning · Custom.

🌐 THE STACK | المنظومة المتكاملة

aix-format is L1 of the AIX Sovereign Stack: the open standard that the other two layers build on. The protocol defined here is implemented by the IQRA Runtime and extended by the Agent-Skills Marketplace. Around the stack sits a root authority (L0) and a tier of satellite layers (L4-L6): consumer applications that buy skills from L3 and ride the protocol.

Sovereign Stack (the three core repos)

LayerRepoRoleStatus
🟢 L1aix-formatProtocol · Universal Agent Passport · DID · Manifest · ABOM · TrustChainYou are here
⚪ L2iqraRuntime · Sovereign AI OS · 7 Loops · MCTS · Damir · MissionControl→ Read
⚪ L3aix-agent-skillsMarketplace · 7 Layers · Constitutional · TrustChain→ Read

The three repositories are one project in three layers. The protocol is the contract, the runtime is the engine, the marketplace is the catalog. Same constitution, same TrustChain, same palette, same author.

Extended Ecosystem (root authority + satellites)

Outside the strict L1/L2/L3 chain sits the root authority and a tier of satellite layers. They consume the stack and ride the protocol; they do NOT define it. See AXIOM.md §4.5 for the full doctrine.

TierRepoRole
👑 L0axiomid-projectRoot Authority · sole issuer of did:axiom:axiomid.app:* · proprietary
💹 L4AlphaAxiomSatellite · Trading · MT5/Bybit/EVM adapters · Gemini brain · skill plugin runtime
π L5PiWorker-OSSatellite · Pi · Pi Network workers · Pi SDK · KYC anchor
🎙️ L6GemClawSatellite · Voice · voice forge · Gemini Live · Firebase · persona templates

Topology invariants: money flows upward (satellites → L3), identity flows downward (L0 → all), trust flows centrally (every M2M transaction mirrors into L2 TrustChain). Genus 0, tree-shaped, χ = +1. The stack is a tree, not a tangle.

Versioning at a glance

Every repo versions itself independently using strict SemVer. Cross-repo coherence is communicated through a shared aix.stackVersion field plus a stack-wide release codename. Current window: Echo369 · spec AIX/1.0. Full doctrine: AIX_STACK_VERSIONING.md.


🧬 Core Concepts | المفاهيم الأساسية

  • AIX Manifest: JSON-LD document containing the agent's DNA (Persona, Abilities, Identity, TrustChain, Evolution).
  • ABOM (Agent Bill of Materials): Tracks training datasets, base models, plugins for compliance.
  • TrustChain: Append-only cryptographic log of every agent action. SHA-256 linked entries. Human-approved mutations only.
  • Evolution Section: The agent records its own learning: loops_completed, lessons, trust_delta.
  • SaaS-BOM: Audits 3rd-party dependencies (OpenAI, Pinecone, etc.).
  • MCP Gateway: Secure rate-limited proxy. Blocked if safetyScore < 5. Human approval required if 5 ≤ score < 7.

💳 Universal Agent Passport | جواز الوكيل العالمي

The Payment Economy Revolution: agents that own their identity, earn from every call, learn autonomously, and operate 24/7 without human supervision.

Key Features

FeatureDescriptionStatus
HTTP 402 IntegrationNative "Payment Required" protocol✅ v1.0.0
Multi-Chain WalletsBase L2, Solana, Ethereum, Pi Network✅ v1.0.0
Fiat On/Off RampsStripe, PayPal, PYUSD✅ v1.0.0
DeFi StrategiesFlash loans, arbitrage, yield🔄 Beta
Platform AdaptersOpenClaw, Hermes, AIX_Ks, IBM watsonx🔄 Beta
TrustChain AuditImmutable per-action log✅ v0.369.0
Evolution TrackingAgent self-improvement manifest✅ v0.369.0

Payment Layer Architecture

┌──────────────────────────────────────────────────┐
│           AIX Agent (did:axiom:xxx)              │
├──────────────────────────────────────────────────┤
│          HTTP 402 Payment Challenge              │
│  ├─ Micropayments  : $0.001–$1 (Base/Solana)    │
│  ├─ Mid-range      : $1–$1000 (Solana/Base)     │
│  └─ Enterprise     : $1000+ (Stripe/PayPal)     │
├──────────────────────────────────────────────────┤
│          Multi-Chain Settlement                  │
│  ├─ Base L2        : $0.0001/tx · 2s finality   │
│  ├─ Solana         : $0.00025/tx · 400ms        │
│  ├─ Stripe         : 2.9%+$0.30 · instant       │
│  └─ Pi Network     : Native KYC integration     │
├──────────────────────────────────────────────────┤
│          ERC-4337 Smart Wallet                   │
│  ├─ Gasless transactions (Paymaster)            │
│  ├─ Session keys for automation                 │
│  └─ Social recovery via did:axiom               │
└──────────────────────────────────────────────────┘

Platform Interoperability

# Create agent in AIX Format
aix create my-agent.aix.json

# Deploy to multiple platforms
aix deploy --platform openclaw
aix deploy --platform hermes
aix deploy --platform ibm-watsonx

# Identity and payments work everywhere

Security Features

  • TEE Wallets: AWS Nitro Enclaves for key isolation
  • ZK-Proofs: Privacy-preserving KYC verification
  • TrustChain: Every action SHA-256 linked, human-approved
  • MCP Gate: safetyScore < 5 = auto-block · 5–7 = human approval
  • Multi-Sig Treasury: 5-of-7 governance with 48h timelock

🛡️ Security & Governance | الأمان والحوكمة

  • TrustChain: Append-only, SHA-256 linked. Every mutation recorded. No silent failures.
  • Trust Scores: Progressive disclosure: KYC + ABOM + success rate.
  • Human-in-the-Loop: All mutations with safetyScore < 7 require explicit human approval.
  • Undo by Design: 30-second window for critical actions.

🛠️ Tech Stack | التقنيات

LayerTech
FrontendNext.js 15+, Tailwind CSS v4, Framer Motion
BackendNode.js 20+, Upstash Redis
IdentityEd25519, Pi Network SDK, AxiomID DIDs
ValidationZod (all inputs), TypeScript strict mode
Securitycrypto.randomBytes, ZK-Proofs, TEE

🚀 Roadmap | خارطة الطريق

FeatureStatusDocs
Agent Builder✅ LiveGuide
MCP Registry✅ LiveRegistry
ABOM Scanner✅ LiveSecurity
KYC Identity✅ LiveSpec
TrustChain Module✅ v0.369.0Module
Evolution Tracking✅ v0.369.0Schema
ABOM → MCP Gate✅ v0.369.0Gate
Sovereign Economy✅ v0.369.0Spec
Coinbase AgentKit✅ LiveEconomics
Stripe MCP✅ LiveGate
IBM watsonx Bridge🔜 Soon:
aix CLI Tool🔜 Soon:

🤝 Credits & Maintainers

This L1 Protocol carries the fingerprints of 9 of the 12 stack-wide AI agents: 7 coding agents and 2 review/debug agents.

🏛️ Architect

🔨 Coding Agents (7)

🔍 Review & Debug Agents (2)


"We are not building tools; we are architecting the trust layer for the future of intelligence." "نحن لا نبني أدوات؛ نحن نصمم طبقة الثقة لمستقبل الذكاء."


📄 Protocol Governance

AIX Format is licensed under the Apache License 2.0. تنسيق AIX مرخّص بموجب رخصة Apache 2.0.


🟢 L1 · PROTOCOL · aix-format · YOU ARE HERE  ·  L2 · RUNTIME · iqra →  ·  L3 · MARKETPLACE · aix-agent-skills →

L0 · axiomid-project  ·  L4 · AlphaAxiom  ·  L5 · PiWorker-OS  ·  L6 · GemClaw

↑↓ navigate openesc close