Skip to main content

PIRΛTHΞΘN

Chafiq El Ibrahimi

Systems-oriented developer and ethical security researcher building privacy-respecting tooling, self-hosted agentic AI, and low-level automation.

Scroll
About Me

Systems-oriented developer focused on privacy and user control

I build software that prioritizes user autonomy, transparency, and security while maintaining high standards of technical excellence and craftsmanship.

I'm a systems-oriented developer and ethical security researcher with a passion for creating technology that serves users rather than platforms. My work focuses on building privacy-respecting tooling and self-hosted agentic AI systems that give users true control over their digital experiences.

My approach to development is guided by several core principles:

Linux-first

I design tools for Unix-like environments first, leveraging native capabilities and POSIX standards rather than relying on cross-platform abstractions that dilute functionality.

Security-aware

Privacy and system integrity are foundational to my work. Encryption, sandboxing, and minimal attack surfaces are integrated from the start, not added as afterthoughts.

System-oriented

I focus on understanding the underlying mechanics—process management, I/O, memory, and networking—rather than stopping at abstracted frameworks that hide important details.

Control over convenience

User autonomy and transparency are paramount. Every action a tool takes should be inspectable, understandable, and reversible when needed.

When I'm not coding, I contribute to open-source security projects and explore low-level system interactions to better understand how to build more resilient and user-respecting technology.

Flagship project

An assistant that lives on your machine — not someone else's cloud.

rika-agent is a persistent, self-hosted agentic AI system that integrates with your shell, file system, and network — built for developers who want control, privacy, and extensibility over a hosted black box.

Function calling
JSON function calling against native provider APIs — no middleware lock-in.
Monitoring
Background system monitoring with zero-token idle; alerts only when something needs attention.
Sandboxing
Three-tier code execution sandbox, from RestrictedPython up to full Docker isolation.
Key storage
AES-256-GCM encrypted key storage at rest.
Interface
Native Telegram UX with live feedback as actions run.
Inference
Ollama and G4F support for local or zero-cost model access.
9:41
A
Aria bot
Selected work

Other tools I've built.

Terminal utilities, desktop applications, and browser tools — built for control and technical depth rather than convenience.

C++ · TUI / GUI

ori

Terminal-first AI assistant with both TUI and web-based GUI. Integrates OpenRouter for multi-LLM access — built for developers who want AI workflows in the terminal or browser.

Python · GTK4

aion-dl

A polished GTK4 + Libadwaita GUI frontend for yt-dlp — per-job controls, concurrent downloads, advanced format and post-processing, subtitle management, and SponsorBlock integration.

JavaScript · Steganography

Harpocrates

Client-side web application for encrypting and hiding files inside images using AES-256-GCM and PBKDF2. Every operation runs locally in-browser — no server uploads, no data leakage.

C++ · CLI

clpbrd

A minimal, cross-platform command-line clipboard utility written in C++ — reliable copy-to-clipboard for scripts and terminal workflows, with no external dependencies.

How I build

Principles, not frameworks.

Linux-first

Tools are designed for Unix-like environments first — leaning on native capabilities, POSIX standards, and terminal-centric workflows rather than cross-platform compromises.

Security-aware

Privacy and system integrity are non-negotiable. Encryption, sandboxing, and minimal attack surfaces are built in from the start, not bolted on after.

System-oriented

Focus on low-level understanding — process management, I/O, memory, and networking — rather than stopping at abstracted frameworks.

Control over convenience

User autonomy and transparency take precedence over opaque automation. Every action a tool takes should be inspectable and reversible.

Get in touch

Open to collaboration on systems, security, and agentic AI.

chafiq.dev@gmail.com

Available for open-source collaboration, security research, and systems-level consulting. Response within 24–48 h.

@piratheon on GitHub @piratheon on Telegram