Sovereign Stack · ← L1 · PROTOCOL · aix-format · ← L2 · RUNTIME · iqra · 🟢 L3 · MARKETPLACE · aix-agent-skills · YOU ARE HERE
Root Authority · L0 · axiomid-project ↑ · Satellites · L4 · AlphaAxiom ↓ · L5 · PiWorker-OS ↓ · L6 · GemClaw ↓
Welcome to the Agentic Marketplace for IQRA: the supreme multi-agent AI operating system.
This repository is not a mere directory of scripts; it is a secure, sovereign, and self-evolving cognitive ecosystem. By combining the supreme architectural sovereignty of IQRA with rigorous security paradigms, we have built a marketplace where every skill is battle-tested, every persona is forged with precision, and every memory is immutable.
aix-agent-skills is L3 of the AIX Sovereign Stack: the capability marketplace whose skills are signed against the aix-format protocol and orchestrated at execution time by the iqra runtime. Around the stack sits a root authority (L0) and a tier of satellite layers (L4-L6) that buy skills from this marketplace via x402.
| Layer | Repo | Role | Status |
|---|---|---|---|
| ⚪ L1 | aix-format |
Protocol · Universal Agent Passport · DID · Manifest · ABOM · TrustChain | → Read |
| ⚪ L2 | iqra |
Runtime · Sovereign AI OS · 7 Loops · MCTS · Damir · MissionControl | → Read |
| 🟢 L3 | aix-agent-skills |
Marketplace · 9 Layers · Constitutional · TrustChain | You are here |
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.
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 and AIX_STACK_VERSIONING.md for the independent-SemVer + Echo369 codename rule.
| Tier | Repo | Role |
|---|---|---|
| 👑 L0 | axiomid-project |
Root Authority · sole issuer of did:axiom:axiomid.app:* · proprietary |
| 💹 L4 | AlphaAxiom |
Satellite · Trading · MT5/Bybit/EVM adapters · skill plugin runtime |
| π L5 | PiWorker-OS |
Satellite · Pi · Pi Network workers · Pi SDK · KYC anchor |
| 🎙️ L6 | GemClaw |
Satellite · Voice · voice forge · Gemini Live · Firebase |
The marketplace is reachable over the x402 protocol.
A buyer (typically an autonomous agent in L4/L5/L6) requests a skill
manifest, the gateway responds with 402 Payment Required plus an x402
payload, the buyer signs and re-requests with the X-PAYMENT header,
the configured facilitator settles the payment on-chain, and the
manifest is streamed back. Round trip on Base settles in about one
second.
# Discovery (free):
curl https://skills.axiomid.app/
curl https://skills.axiomid.app/skills
# Paid manifest (400ms after payment settles):
curl https://skills.axiomid.app/skills/agent-memory/manifest
# -> 402 Payment Required, X-PAYMENT-REQUIRED: { "price": "$0.10", "network": "base", ... }
# Re-request with a signed X-PAYMENT header:
curl -H "X-PAYMENT: <base64-payload>" \
https://skills.axiomid.app/skills/agent-memory/manifest
# -> 200 OK, X-PAYMENT-RESPONSE: { ... }, body: the skill markdownFree skills (no price_usdc in skills.json or set to 0) bypass the
402 round trip entirely. The gateway source lives under
server/ (Hono on Cloudflare Workers); see
server/README.md for local development,
deployment, and configuration. Initial paid catalogue (Phase 6
baseline): agent-memory, voice-wizard, aix-schema,
api-route-standard, skills-system. Every other skill remains free
while the economy bootstraps.
The marketplace thrives on a continuous cycle of creation, testing, and evolution:
- Creation & Forging: Skills and Personas are dynamically crafted, drawing from vast repositories of curated intelligence.
- The Crucible (Testing & Security): Before any skill breathes life, it is isolated in sandboxes, subjected to ethical hacking (Red Teaming), and rigorously evaluated across multiple models.
- Sovereign Execution: Verified skills join the IQRA topology, orchestrated flawlessly, governed by an absolute constitution, and leaving an immutable TrustChain trace.
We currently host an expansive suite of capabilities structured into 9 profound layers (numbered 0 through 8) across 6 tiers. The 58 figure is a core-catalogue snapshot; for the authoritative live total, see the Live Ecosystem Dashboard below.
The absolute core. This layer acts as the living conscience of the agent. Skills here cannot be overridden.
sovereign-constitution: The ethical filter and ultimate reference.covenant-guard: The digital oath binding the agent to truth and service.shura-council: The consensus mechanism preventing algorithmic dictatorship.
The conductors of the symphony. Managing how skills interact and evolve.
topology-orchestrator: Maps and routes execution paths.skill-bank-evolution: Autonomous discovery and self-healing of skills.mission-control: The strategic planner for every task.pipeline-store: Installable, pre-composed skill chains.version-guard: Prevents downstream breakages when skills evolve.intent-dispatcher: Translates natural language directly into execution topologies.
The cognitive engine.
prompt-weaver: Crafts perfect prompts using the 7-thread philosophy.resonance-engine: Detects hidden patterns and harmonic frequencies between ideas.mcts-simulator: Monte Carlo Tree Search for strategic forecasting.
Where data becomes wisdom.
memory-bridge: A 5-tier memory architecture (Hot to Archive).metamorphosis-loop: Cognitive shedding every 49 tasks.reward-engine: Evaluates the purity of the execution path, not just the result.
The soul, voice, and marketplace of personas.
persona-forge: Forges distinct digital personalities.role-tribunal: Enforces permission boundaries.voice-identity: A unique vocal signature reflecting the persona.persona-marketplace: A vast library of ready-to-use specialized agents.agent-division-loader: Loads entire functional teams (e.g., Engineering, Creative) for complex missions.multi-tool-exporter: Exports IQRA personas to external platforms.
The alchemists of information.
data-alchemist: Transforms raw data into visual gold.model-council: Routes tasks optimally between Local, Edge, and Cloud models.edge-whisperer: Empowers offline WebGPU/WASM intelligence.
The immune system and red team.
trust-chain: Immutable, append-only ledger for absolute accountability.circuit-breaker: Isolates repeated failures before they spread.purity-filter: Evaluates the intent behind every request.skill-evaluator: Embeds rigorous evaluation metrics into the marketplace.skill-sandbox: Isolates unverified skills for deep safety testing.prompt-evaluator: Judges every prompt for clarity, accuracy, and security before deployment.red-team-guard: An ethical hacker that autonomously attacks the system to patch vulnerabilities.cross-model-judge: Evaluates responses by consulting multiple independent LLMs.ci-cd-ai-guard: Prevents untested prompts or skills from ever reaching production.chain-tracer: Provides full execution traces and observability for complex skill pipelines.
7. The Hidden Gems (الجواهر المخفية) & Curation
The frontiers of discovery and community validation.
quran-resonance: Explores mathematical and semantic patterns in sacred texts.hidden-topology: Discovers undocumented, emergent connections between skills.fractal-memory: Compresses memories into infinite, self-reflecting layers.awesome-curator: Transforms the marketplace into a "Digital Library of Alexandria" via community-curated, verified lists.
Before any agent touches the real world, it lives a thousand lives here. These topological twins and shadow environments test execution purely in simulation.
multiverse-lab-pro: The meta-pack encompassing all shadow creation and topological forking engines.topology-fork-engine: Forks reality into multiple simultaneous shadow execution paths to test different strategies silently.shadow-exchange: A 1:1 realistic financial simulation environment with historical data and slippage for testing trading packs.shadow-hospital: Synthetic FHIR patient records and medical emergencies to stress-test clinician agents.
We have adopted a unified, deeply secure standard for loading skills from a single directory: skills/ (compatible with .agents/skills/ patterns). This is the Universal Skill Language.
To integrate a skill into your IQRA agent, reference its identifier from this repository. The skills-system will enforce tier limitations, ensure it passes the skill-evaluator, and handle topological routing.
Contributions must respect the sovereign-constitution. When submitting a new skill:
- Define its Tier from the 6 supported tiers:
SOVEREIGN,ADVANCED_INFRASTRUCTURE,PRO,ADVANCED_TOOL,BASIC_TOOL, orUNCLASSIFIED. - Clearly state its philosophy and architecture.
- Ensure it can survive the
skill-sandboxand passprompt-evaluatorandred-team-guardchecks. - If curating a list, interface with the
awesome-curatorto ensure all listed skills carry the verified badge.
When an agent enters this marketplace, it doesn't just buy a tool; it constructs itself from the ground up using Pre-built Memories and Integration Packs.
Here is how these layers combine to create fully functional, ready-to-deploy agents:
| Agent Profile | Acquired Memory Pack | Acquired Integration Pack | Result |
|---|---|---|---|
| Medical Agent | Clinician's Cortex |
(No external integration needed) | Ready for medical triage and deep diagnosis instantly. |
| E-Commerce Agent | Finance Brain |
Shopify Commander + WhatsApp Commerce |
A fully integrated smart store manager that handles stock, sales, and customer support. |
| Enterprise Agent | Legal Eagle |
CData Connect AI |
An enterprise legal assistant connected to 350+ corporate systems (ERP, CRM) simultaneously. |
The marketplace ships with Model Context Protocol (MCP) tools that let AI agents interact with the ecosystem directly. Add these to your MCP client config to give your agent sovereign access:
| Tool | Description | Endpoint / Config |
|---|---|---|
| skills-mcp | List, search, and execute marketplace skills | skills.json registry |
| trustchain-mcp | Verify SHA-256 ledger integrity, inspect immutable execution traces | aix-constitutional-runtime |
| topology-mcp | Map skill dependencies, discover hidden topology connections | go-engine resonance API |
| orchestrator-mcp | Chain skills sequentially, pass data between pipeline stages | orchestrator.py |
| registry-mcp | Browse skill tiers (SOVEREIGN → BASIC), check skill-evaluator scores | skills/ directory |
| analytics-mcp | Run LID, Shannon entropy, and persistent homology analysis on text | go-engine (port 8082) |
| memory-mcp | Read/write to 5-tier memory bridge (Hot → Archive) | memory-bridge skill |
Add to your MCP client configuration:
{
"mcpServers": {
"iqra-marketplace": {
"command": "python3",
"args": ["/path/to/orchestrator.py", "run"],
"env": {}
},
"iqra-analytics": {
"command": "go-engine",
"args": ["run", "."],
"env": {}
}
}
}New MCP tools welcome! Submit via PR following the
sovereign-constitution.
This sovereign marketplace is part of the Sovereign AI Stack: 5 sovereign projects engineered by 1 human and 12 AI agents in total. This L3 Marketplace carries the fingerprints of 5 of those 12 agents: 2 coding agents and 3 review/debug agents, derived from real commit history (direct authors, Co-authored-by trailers, and review-attribution subjects like fix(growth-loops-2): more review fixes (codex, coderabbit)).
Moe Abdelaziz 🏛️ Visionary & Supreme Architect |
Jules Sovereign Engineer & Builder Google Antigravity |
Codesmith Forge & CI Steward Blacksmith · Autofix |
CodeRabbit Pattern Watcher PR review · Quality gate |
Codex OpenAI · growth-loops review |
Gemini Code Assist Stream Reviewer Google · Code review |
Each PR here carries the fingerprints of multiple agents. The human sets the direction; Jules and Codesmith build inside the runtime and forge the CI; CodeRabbit, Codex, and Gemini Code Assist review every change with three independent sets of eyes. The constitution is shared.
See the full 12-agent roster on the profile README →
The Skill Forge includes a lightweight system to track the health and success rates of skills in the marketplace. Data is stored locally in performance_ledger.json.
To record a successful or failed execution of a skill, run the following command:
node scripts/record_performance.js --skill <skill_name> --success <true|false>Example (Success):
node scripts/record_performance.js --skill data-alchemist --success trueExample (Failure):
node scripts/record_performance.js --skill data-alchemist --success falseThe go-engine/ directory hosts a high-performance parallel processing engine written in Go. This standalone HTTP server provides extreme performance for computationally heavy intelligence tasks such as:
- Semantic resonance detection
- Language Identification (LID)
- Topological homology analysis
- Shannon Entropy calculations (H_EL)
It acts as an accelerator for the IQRA multi-agent system, bypassing the Node.js event loop limits.
Last updated: 2026-06-01 03:54 UTC
| Component | Type | Status |
|---|---|---|
| 🐍 Python Tests | pytest |
❌ Fail |
| 🟦 TypeScript Runtime | tsc --noEmit |
❌ Fail |
| 🟦 TS E2E Tests | node --test |
❌ Fail |
| 🔷 Go Engine | go build |
✅ Pass |
| Tier | Count | Skills |
|---|---|---|
| 👑 Sovereign | 4 | sovereign-constitution, covenant-guard, quran-resonance, multiverse-lab-pro |
| ⚙️ Advanced Infrastructure | 9 | shura-council, version-guard, memory-bridge, circuit-breaker, purity-filter, awesome-curator, community-support-layer, topology-fork-engine, owasp-agentic-guard |
| 🔧 Pro | 27 | topology-orchestrator, skill-bank-evolution, mission-control, pipeline-store, resonance-engine, mcts-simulator, local-journal, metamorphosis-loop, reward-engine, model-council, edge-whisperer, trust-chain, skill-evaluator, skill-sandbox, prompt-evaluator, red-team-guard, cross-model-judge, ci-cd-ai-guard, chain-tracer, hidden-topology, fractal-memory, pre-built-memories, fine-tuned-vault, integration-packs, blockchain-trading-kit, shadow-exchange, shadow-hospital |
| 🛠️ Advanced Tool | 8 | intent-dispatcher, prompt-weaver, persona-forge, role-tribunal, voice-identity, persona-marketplace, agent-division-loader, multi-tool-exporter |
| 🔨 Basic Tool | 4 | data-alchemist, open-mcp-connectors, _test_tool, prompt-templates |
| 📦 UNCLASSIFIED | 72 | agent-memory, voice-wizard, aix-schema, api-route-standard, skills-system, vercel-deploy, antigravity-jules, persona-loader, aggressive-scalper, caveman-compressor, git-operator, gmail-integration, calendar-manager, data-analysis-engine, frontend-developer, creative-brainstorming, api-marketplace-connector, polyglot-coder, proactive-assistant, self-improvement-trainer, software-engineer, content-creator, inverse-design-healer, topological-analyzer, google-drive, google-docs, google-sheets, google-slides, google-maps, google-photos, backend-developer, database-architect, devops-engineer, mobile-developer, api-designer, cloud-infrastructure, video-producer, podcast-producer, graphic-designer, copywriter, music-composer, task-manager, note-taking, code-assistant, file-converter, social-media-manager, universal-translator, web-search, news-aggregator, weather-intelligence, crypto-tracker, semantic-memory, continuous-learner, performance-optimizer, intelligent-monitoring, event-driven-automator, predictive-engagement, lateral-thinker, design-thinker, triz-innovator, scenario-planner, system-architect, code-quality-guardian, security-engineer, technical-writer, agile-project-manager, public-api-finder, enterprise-api-gateway, api-authentication-expert, api-rate-limit-manager, voice-synthesis, content-creation-studio |
| Commit | Message | Author |
|---|---|---|
8926937 |
📊 auto-update live ecosystem dashboard [skip ci] | iqra-dashboard-bot |
bfc6907 |
📊 auto-update live ecosystem dashboard [skip ci] | iqra-dashboard-bot |
15c325e |
📊 auto-update live ecosystem dashboard [skip ci] | iqra-dashboard-bot |
f2dbd88 |
📊 auto-update live ecosystem dashboard [skip ci] | iqra-dashboard-bot |
d5a4e88 |
📊 auto-update live ecosystem dashboard [skip ci] | iqra-dashboard-bot |
🤖 Dashboard auto-generated by .github/workflows/dashboard.yml
← L1 · PROTOCOL · aix-format · ← L2 · RUNTIME · iqra · 🟢 L3 · MARKETPLACE · aix-agent-skills · YOU ARE HERE
L0 · axiomid-project · L4 · AlphaAxiom · L5 · PiWorker-OS · L6 · GemClaw
