# Ouail Bni (ProxySoul)

> Personal site of Ouail Bni — senior software engineer & tech lead at Axis Communications (Sweden) and agent harness engineer. Creator of Empryo, a graph-powered AI coding agent that edits code by symbol (AST + LSP) instead of string matching, and of its predecessor SoulForge (840+ GitHub stars).

Key facts:
- Specialty: agent harness engineering — tool design, context management and compaction, prompt caching, multi-model routing, sandboxed execution, and evaluation loops for LLM agents.
- Builds Empryo (https://empryo.com), the AI coding agent with a live code "genome" (ranked dependency graph). GitHub home: https://github.com/proxysoul/soulforge
- Previously built SoulForge (https://github.com/proxysoul/soulforge), the v2 experiment that proved graph-powered code intelligence beats grep-driven agents.
- 6+ years full-stack: TypeScript, React, Next.js, Node, React Native. M.Sc. in Computer Science, Malmö University.

## Pages

- [Home](https://proxysoul.com): who I am and what I'm building now
- [About](https://proxysoul.com/about): background, experience timeline, and philosophy
- [Projects](https://proxysoul.com/projects): Empryo, SoulForge, Jurict, and my open-source GitHub repositories
- [Tech Stack](https://proxysoul.com/stack): tools and technologies, including the AI & agent harness engineering toolkit
- [Blog](https://proxysoul.com/blog): technical deep-dives on AI agents, code intelligence, and engineering

## Blog posts

- [Agent Harness Engineering: The Discipline That Decides How Good Your AI Agent Is](https://proxysoul.com/blog/agent-harness-engineering): Agent harness engineering is the practice of designing everything around the LLM — tools, context, caching, routing, evaluation. It's why the same model can be brilliant in one product and useless in another. A field guide from building Empryo and SoulForge.
- [OpenCode vs Claude Code vs SoulForge: a real comparison](https://proxysoul.com/blog/opencode-vs-claude-code-vs-soulforge): Three terminal coding agents, three philosophies. Provider freedom, vertical Anthropic stack, or graph-powered code intelligence. Honest matrix, source-verified facts, zero marketing fog.
- [I built an AST editor for AI because string matching is killing your codebase](https://proxysoul.com/blog/ast-editor-for-ai): Every AI coding tool in 2026 still edits your files with str_replace. I rebuilt the editing primitive on the TypeScript compiler API. Here's what changes when the agent treats code as structure, not text.
- [Introducing SoulForge: Graph-Powered Code Intelligence](https://proxysoul.com/blog/soulforge): I wanted to trust AI coding agents. That meant building one that understands the codebase as a system, not file by file, but as a graph with structure, coupling, and consequences.
- [What I Learned Building JuriCt](https://proxysoul.com/blog/building-jurict): Lessons from building a live AI deliberation platform — real-time orchestration, prompt engineering for personality, and why the hardest part of AI apps isn't the AI.
- [From Mr-wii to ProxySoul: 7 Years of Portfolio Evolution](https://proxysoul.com/blog/evolution): A look back at how my portfolio evolved across three major versions — from a static HTML page in 2019 to a terminal-inspired Next.js app in 2026.
- [Hello, ProxySoul](https://proxysoul.com/blog/hello-proxysoul): Welcome to my rebranded portfolio. From Pouiiro to ProxySoul — a fresh start with the same passion for building great software.

## Full content

- [llms-full.txt](https://proxysoul.com/llms-full.txt): complete markdown content of every blog post
- [RSS feed](https://proxysoul.com/feed.xml)

## Contact

- GitHub: https://github.com/proxysoul
- LinkedIn: https://linkedin.com/in/ouailbni
- X: https://x.com/BniWael