Ecosystem Pillar 02


Not a fragile extension. Not a bolt-on chatbot. A fully native, local-first AI IDE with embedded Rust GPU inference, Machine HealthGuard thermal safety, and an integrated visual Motion Studio.

Built for Agents.
Engineered for You.

Fearless Refactoring in the Shadows.

Stop worrying about AI breaking your codebase. Astrolabe executes complex multi-file edits inside a secure, hidden shadow workspace. Review the diffs, test the logic, and click "Approve." Your root files are only touched when you say so.

A Memory That Never Fades.

Astrolabe doesn't just read your active tab. The built-in semantic graph deeply analyzes and maps your entire workspace context instantly. The agent remembers architectural decisions and file relationships across sessions natively.

Your Code. Your Keys.

Completely stripped of corporate surveillance. Bring your own secure keys, or use our built-in offline routing models. Zero middleman servers reading your proprietary algorithms.

More Than Just
An Editor.

It's An Operations Center.

1-Click Cloud Hosting Deployments

Why leave the IDE to deploy? Astrolabe features a native Hosting Console. Compile your static bundles, spin up isolated containers, and distribute to the Edge seamlessly directly from your sidebar.

Hyper-Fast Ghost Text

Experience real-time, ultra-fast inline code completions powered by our optimized prediction engine. Use cloud intelligence or spin up local models for instantaneous suggestions.

Deep Terminal Integration

Astrolabe's AI natively reads your real-time linter diagnostics, error matrix, and terminal outputs to suggest instantaneous fixes.

Autonomous Mode & Web Search

Toggle "Autopilot Mode" to let the agent run iterative terminal commands, scrape active documentation from the web, and solve complex bugs completely hands-free.

Phase 1: The Request

You speak in intent. Astrolabe parses the goal.

Phase 2: The Blueprint

The Semantic Graph constructs an architectural map of changes.

Phase 3: The Execution

Parallel multi-file generation inside the shadow workspace.

Phase 4: The Merge

Flawless execution. Verify native diffs and commit.

Astrolabe Workspace
1 // Auth module initialized.
2 export function verifyToken() {
3 return "deprecated_v3";
4 }
Migrate auth module to v4 across the entire workspace.
Semantic Architecture Plan
14 Files Impacted
Shadow Execution in Progress...
1 // v4 authentication handler
+ import { OAuth2Client } from '@v4/auth';
+ import { GraphCache } from '@v4/cache';
4 export async function verifyToken(req: Request) {
- const token = req.headers.get("Auth");
+ const client = new OAuth2Client(process.env.KEY);
+ const session = await client.verify(req);
+ await GraphCache.update(session.id);
+ return session;
10 }
11 // Execution proceeding to router.ts...

Changes Safely Merged

14 Files updated. 0 conflicts. 0 CI failures.

Ready for deployment

Architectural Comparison

How native all-in-one local AI architecture compares against legacy IDE and extension configurations.

Capability / ArchitectureAstrolabe IDE (All-In-One Native)Current Market Standards (Legacy IDEs + Extensions)
Local Inference Engine✅ Embedded Rust Daemon (Zero external setup)❌ Requires installing & managing separate background CLI tools (Ollama, LM Studio)
Hardware & Thermal Protection✅ Machine HealthGuard (Real-time thermal & VRAM guard)❌ None (Risk of thermal throttling and unhandled OOM panics)
Model Management✅ Integrated Hugging Face Hub (1-click GGUF download)❌ Disconnected. Manual terminal downloads & port configuration
Visual UI & Motion Studio✅ Astrolabe Motion Studio (Live 3D & GSAP timeline compiler)❌ None (Plain text editor only)
Offline & Air-Gapped Privacy✅ 100% Offline & Sovereign (Zero cloud telemetry)⚠️ Cloud-dependent or fragmented local plugin configurations
Subscription Traps✅ 100% Free & Open Source (MIT License)❌ $20/month proprietary cloud subscriptions
Extension Ecosystem✅ Full VS Code Extension Compatibility⚠️ Often limited custom extension APIs
Codebase Intelligence✅ Tree-Sitter AST + Semantic Graph DAG + Local ONNX Vectors❌ Flat regex search or cloud-dependent remote indexing
Universal Local API Gateway✅ Built-in OpenAI (/v1) & Anthropic endpoints on port 4040❌ Closed proprietary cloud APIs or fragmented CLI daemons
Intelligence Core

Tree-Sitter AST, Semantic Graphs & Local Vectors

Astrolabe does not treat your codebase as flat text chunks. It compiles exact abstract syntax trees, maps project-wide call hierarchies, and measures the blast radius before the agent proposes a single diff.

WebAssembly Tree-Sitter AST

Extracts compiler-grade syntax trees for TypeScript, JavaScript, Python, Rust, Go, and C++. Classes, decorators, methods, and variable scopes are indexed as structured language symbols rather than naive regex lines.

Topological Semantic Graph

Constructs an in-memory Directed Acyclic Graph (DAG) mapping caller-callee relationships across every file. Automatically runs blast-radius impact analysis so complex refactors never break downstream code.

Hybrid Vector + BM25 Engine

Combines local ONNX vector embeddings (BGE-Small / MiniLM) with BM25 lexical keyword matching via Reciprocal Rank Fusion (RRF), ensuring both high-level semantic intent and exact symbol names are retrieved accurately.

0.207s
Extension Host Activation
< 0.8s
Cold Window Launch
~140 MB
Base RAM Footprint
0 KB
Runtime Node Modules

System & GPU Architecture

Astrolabe is engineered to scale dynamically from lightweight APU laptops to multi-GPU workstations.

Tier 1: Lightweight & Integrated Graphics (1B - 3B Models)

Optimal for quick local completions and coding assistance on ultrabooks and APUs.

Minimum Specification

Operating SystemLinux (Ubuntu 20.04+, Arch, Debian) / Win 10/11 x64
Processor / CPUIntel Core i5 (8th Gen+) / AMD Ryzen 5 / AVX2
GPU AcceleratorAMD Radeon 680M / Intel Iris Xe / Vulkan 1.3
Memory & VRAM8GB RAM (4GB Shared/Dedicated VRAM)
Target ModelsGemma 3 1B/3B, Qwen 2.5 1.5B (Q4/Q8)

Recommended Specification

Operating SystemLinux / Windows 11 (Vulkan / SGLang enabled)
Processor / CPUAMD Ryzen 7 (7000/8000 series) / Intel Core i7
GPU AcceleratorAMD Radeon 780M / Intel Arc A380 (Vulkan 1.3)
Memory & VRAM16GB RAM (6GB-8GB VRAM allocation)
Target ModelsQwen 2.5 Coder 3B, TinyMistral (Q8_0)

Tier 2: Dedicated GPU & Workstation (7B - 32B+ Agentic Models)

Full local intelligence, multi-file agentic code synthesis, and deep reasoning models.

7B - 14B Mainstream Developer

GPU HardwareAMD RX 6700/7700+, NVIDIA RTX 3060/4060/4070
Dedicated VRAM8GB - 12GB Dedicated GDDR6
System Memory16GB - 32GB System RAM
Acceleration BackendNative Vulkan 1.3 / CUDA Kernels
Target ModelsQwen 2.5 Coder 7B, DeepSeek-Coder 7B (Q4_K_M)

27B - 32B+ Pro Workstation

GPU HardwareNVIDIA RTX 3090 / 4090 / AMD RX 7900 XTX / Pro
Dedicated VRAM16GB - 24GB+ High-Bandwidth VRAM
System Memory32GB - 64GB+ High-Speed RAM
Hardware SafetyMachine HealthGuard (Thermal & OOM Active)
Target ModelsQwen 2.5 Coder 32B, Bonsai 27B, MoE Quants
Open Ecosystem

Universal Local API Gateway

Astrolabe exposes standard OpenAI and Anthropic compatible endpoints on 127.0.0.1:4040. Connect Aider, Continue, Cline, OpenHands, or custom LangChain scripts directly to your local GPU runtime.

Aider CLI Integrationbash

# Direct connection to Astrolabe GPU Daemon

aider --openai-api-base http://127.0.0.1:4040/v1 \

--model openai/astrolabe-local

Continue.dev / Cline Configjson

{

"title": "Astrolabe Local GPU",

"provider": "openai",

"model": "astrolabe-local",

"apiBase": "http://127.0.0.1:4040/v1"

}

Install in Seconds. No bloatware.

🐧 For Linux (Debian/Ubuntu, Arch, Fedora)

bash

# Securely download and install Astrolabe from exovon.in

curl -fsSL https://exovon.in/install.sh | bash

# Launch it instantly

astrolabe .

🪟 For Windows (Windows 10/11 x64)

download

# Direct standalone Windows archive

astrolabe-windows-x64.zip

Download ZIP

Join the Next Era of Software Engineering.

Upgrade to a Native AI IDE with zero-setup local GPU inference, visual Motion Studio, and Machine HealthGuard.