HudHudScript

Agentic Programming Loop Engineering

Goal-based automation with AI agents, governance councils, MCP, and autonomous loops

agentic.hudhud
// AGENTIC PROGRAMMING
// Provider and council topology. No paid API call is made in this offline demo.

provider DeepSeek {
    type: "deepseek";
    model: "deepseek-chat";
    api_key: env("DEEPSEEK_API_KEY");
}

provider OpenAI {
    type: "openai";
    model: "gpt-4o";
    api_key: env("OPENAI_API_KEY");
}

provider Anthropic {
    type: "anthropic";
    model: "claude-3-5-sonnet-20241022";
    api_key: env("ANTHROPIC_API_KEY");
}

provider Kimi {
    type: "kimi";
    model: "moonshot-v1-8k";
    api_key: env("KIMI_API_KEY");
}

agent Planner {
    role: "planner";
    intent: "Turn the question into an evidence-backed plan";
    provider: DeepSeek;
}

agent Critic {
    role: "critic";
    intent: "Find unsupported assumptions and failure modes";
    provider: Anthropic;
}

agent Implementer {
    role: "implementer";
    intent: "Produce the smallest verifiable implementation";
    provider: OpenAI;
}

agent Judge {
    role: "judge";
    intent: "Compare evidence and issue the final verdict";
    provider: Kimi;
}

let council_members = ["Planner", "Critic", "Implementer", "Judge"];
let quorum_required = 3;

council ModelCouncil {
    quorum: quorum_required;
    members: council_members;
    rules: ["cite-evidence", "critic-must-review", "three-votes-required"];
}

let question = "How should we ship a safe parser change?";
let roles = [Planner.role, Critic.role, Implementer.role, Judge.role];
let provider_types = [
    DeepSeek.type,
    Anthropic.type,
    OpenAI.type,
    Kimi.type
];

print("Question: " + question);
print("Council members: " + council_members.length);
print("Council quorum: " + quorum_required);
print("Flow: " + roles[0] + " -> " + roles[1] + " -> " + roles[2] + " -> " + roles[3]);
print("Providers: " + provider_types[0] + ", " + provider_types[1] + ", " + provider_types[2] + ", " + provider_types[3]);

Agentic: 4 providers (DeepSeek, OpenAI, Anthropic, Kimi). ModelCouncil with Planner, Critic, Implementer, Judge agents.

Code in Your Language

Same agentic loop, 23 human languages. Write agents, governance, and automation in your native tongue.

All Programming Paradigms in One Language

HudHudScript uniquely combines agentic, functional, OOP, SOP, governance, and multi-agent paradigms

🔄

Agentic Loop Engineering

Define agents with providers and let them iterate autonomously. Provider-powered loops that run until objectives are met.

🏛️

Governance-First

Constitution, laws, councils, and swarms embedded in the language. Agents operate within governed boundaries.

OO

Object-Oriented

Full OOP: classes, inheritance, polymorphism, encapsulation. Objects with methods and properties in every language.

SO

Subject-Oriented

Subjects, domains, and hierarchical agent organization. SOP primitives for complex agent architectures.

λ

Functional Programming

First-class functions, closures, immutability, and higher-order operations. Compose functions like data.

🪼

Multi-Agent & Swarm

Coordinate multiple agents in parallel or sequential swarms. Communities, broadcasts, and inter-agent messaging.

23

Multi-Lingual

Write code in 23 human languages. Same program runs identically — Arabic, Japanese, Turkish, English, and more.

🔌

MCP + AI Provider Native

DeepSeek, OpenAI, Anthropic, Ollama. MCP connects to 90+ services: PostgreSQL, Redis, Git, Filesystem.

🖥️

Web + TUI + Async

Web module, TUI widgets (split, list, gauge), native async/await, Promise combinators, event-driven architecture.

9
Paradigms
5+
AI Providers
90+
MCP Services
23
Languages

Build Autonomous Agents Today

Define agents, set goals, establish councils — the loop does the rest. Multi-lingual, AI-native, governance-built-in.