Quick Front
A web app that turns a short project brief into a ready frontend starter, guided by an LLM that picks the stack.
Building an LLM project wizard around stacks, features and code.
The challenge
Starting a project shouldn't feel like a week of setup and guessing.
Developers lose hours picking a framework, wiring up tooling and copying configs, before they write a single feature.
The goal was to build a wizard where an LLM asks the right questions, recommends a stack and generates the code.
Understanding the data
Fewer decisions. Faster starts.
Instead of a blank prompt, the wizard walks users through six short, focused steps.
At each step the context so far is packed into a structured prompt so answers stay on topic.
The model
Guided generation for a developer tool.
GigaChat answers are rate-limited and cached in Redis, while a generator service turns the plan into a downloadable template.
GitHub integration can push the generated project straight into a new repository, ready to run.
The model
Guided generation for a developer tool.
GigaChat answers are rate-limited and cached in Redis, while a generator service turns the plan into a downloadable template.
GitHub integration can push the generated project straight into a new repository, ready to run.
Pipeline
Ready
by design.
The final system brings
LLM guidance, code generation
and GitHub integration
into one coherent experience.
The main engineering rule remained
consistent throughout the project:
save setup time without hiding
choices.