Hi again, Flet community!

In my previous post, I introduced fluent-manager — a library I built to simplify Project Fluent integration in Flet apps like CodeUA. Today, I want to share a "Success Story" about taking this a step further: implementing full localisation in a new project without writing a single line of code myself.

The Project: H-Line (Hotline App)

I’ve been working on H-Line, an open-source tool for the Ukrainian public sector to automate citizen appeals. It’s a more complex project using the Registry pattern and a centralized state container.

The AI-Agent Approach

Instead of manual migration, I used a specialized AI Agent Skill. For those using AI-integrated CLI tools (like Gemini CLI), a "Skill" is essentially a package of expert knowledge that teaches an agent exactly how to use a specific library or pattern.

I "taught" my agent the fluent-manager skill, and it autonomously:

1. Analyzed the existing H-Line architecture.
2. Integrated fluent-manager into the state container.
3. Migrated the entire UI from hardcoded strings to .ftl resources.
4. Set up the localization registry in main.py.

How to use this Skill

The beauty of this approach is that you can now give your own AI agents the same "expertise." If you are using a compatible agent system, you can install this skill with one command:

npx skills add BogdanovychA/fluent-manager --skill fluent-manager

Once installed, your agent knows the fluent-manager API, the directory structure it expects, and how to best inject it into Flet's view hierarchy.

Why it matters

By using an AI Skill instead of just raw prompts:

• Architecture Consistency: The agent followed the project's SOLID principles and Registry pattern perfectly.
• Zero Hallucinations: Because the skill contains the actual library documentation and best practices, the agent didn't "guess" the API.
• Speed: The entire migration of a multi-route app took minutes.

Links

• H-Line on GitHub: BogdanovychA/h-line
• fluent-manager Skill: Available via npx skills

Localisation is often the most tedious part of app development. Combining Flet, Project Fluent, and AI Skills makes it almost invisible.

Would love to hear if anyone else is using AI agents to manage their Flet project's architecture!