Death Legion Team (GitHub: deathlegionteamlk) is an open source group from Sri Lanka, building under the name DEMO × HEXA.

We've shipped apps with most of the popular frameworks, ran into the same walls every time, and eventually stopped complaining and started building. That's the whole origin story.

Three projects so far. LEGIONGASPER is a Python multi-agent AI system that coordinates parallel agents using a captain-agent pattern. LEGIONHERCULES is a local CLI agent framework that runs entirely on your own hardware via Ollama — no API keys, no cloud, no billing surprises. legionnative is a cross-platform mobile framework in TypeScript with a Rust core, built because we were tired of 3-second cold starts and bundles that ballooned with every dependency.

All three are MIT licensed. All three are built in public. All three exist because we needed them and nothing out there did the job.

LEGIONGASPER is a Python multi-agent AI system. You give it a task. A captain agent breaks it into subtasks, assembles a squad of specialized agents, runs them in parallel, and merges the results back. You don't wire the handoffs. You don't babysit the coordination.

It runs on FastAPI with a REST + WebSocket API on port 8081, and ships with a live dashboard at localhost:8080 — active agents, task queue, completion rates, cost per provider, all updating live.

Memory runs across five layers so context isn't lost at the wrong time: Session (current run) → Working (current task) → Daily (survives the session) → Long-term (ChromaDB vector store, survives reboots) → Collective (shared across all agents).

# LEGIONGASPER — captain-agent pattern
captain = Captain(agent_id="research_lead")

subtasks = captain.decompose_task("Analyze market trends in AI tooling")
squad    = captain.form_squad(subtasks)
results  = await squad.execute_parallel()
final    = captain.aggregate_results(results)

LEGIONHERCULES is a CLI framework for autonomous AI agents that run 100% locally via Ollama. No API key. No cloud. No billing page. Pull a model once and run agents on your own hardware, free.

These aren't chatbots with a tool call bolted on. They plan multi-step tasks, pick tools, execute them, read the results, and keep looping until they're done. Multiple agents run in parallel. The terminal UI uses Rich; the CLI uses Typer.

# LEGIONHERCULES — running in 3 commands
ollama serve
ollama pull llama3.2
legionhercules chat --agent coder "Refactor main.py to use async/await"

 

legionnative came from shipping apps with Expo, React Native, and Flutter and hitting the same problems every time: 3-second cold starts, hot reload that lied about being instant, bundles that grew 400KB when you added one dependency, and "offline-first" that really meant "cache one screen."

TypeScript runtime. Rust core for what JS is slow at. Vite for dev. Skia for consistent rendering. WASM plugin system so native code can be written in any language.

# legionnative — first app in 4 commands
npm install -g @legion/cli
legion create my-app
cd my-app
legion start

// legionnative — what a component looks like
import { View, Text, Button } from '@legion/runtime';
import { useState } from 'react';

export default function App() {
  const [count, setCount] = useState(0);
  return (
    <View style={{ flex: 1, alignItems: 'center', justifyContent: 'center' }}>
      <Text style={{ fontSize: 24 }}>Count: {count}</Text>
      <Button title="Add One" onPress={() => setCount(count + 1)} />
    </View>
  );
}

Stack: TypeScript Runtime · Rust Core · Node.js CLI · Vite HMR · Skia Rendering · WASM Plugin System

              

Languages

Frameworks & Runtimes

AI & Storage

Infrastructure & Tools

 

All three projects are actively maintained. PRs welcome. Bug reports too — especially ones with reproduction steps.

Check the CONTRIBUTING.md in each repo before opening a PR. Mostly so we don't duplicate work.

         

_{Death Legion Team · deathlegionteamlk · DEMO X HEXA · Sri Lanka open source developers · LEGIONGASPER multi-agent AI system Python FastAPI · LEGIONHERCULES local CLI agents Ollama · legionnative cross-platform mobile framework TypeScript Rust · Death Legion LK GitHub · autonomous AI agents open source · local LLM CLI Python · mobile framework Sri Lanka · Death Legion Team open source projects · deathlegionteamlk GitHub profile}