mk31231293180/Coding_AI

Free Web Senior Engineer AI

This edition is built to be web-hosted instead of trapped on one computer.

It is designed for a free stack:

• —Hugging Face Docker Space for web hosting,
• —Cloudflare Workers AI for the model calls,
• —your browser for the UI,
• —zipped project upload/download for repo transfer.

What it does

• —works like a normal AI assistant,
• —plans before coding,
• —reads repo structure and related files,
• —edits code for React, Python, and Google Apps Script,
• —runs a multi-agent flow: planner, architect, retriever, coder, debugger, reviewer,
• —performs lexical + semantic retrieval,
• —runs validation commands when possible,
• —blocks obvious placeholder output,
• —lets you upload a zipped project and download the modified zip.

Important reality check

This is free to deploy and use within free-tier limits, but it is not infinite free compute.

That means:

• —hosting is on a free CPU tier,
• —the model backend uses a free daily quota,
• —storage is ephemeral unless you pay for persistence,
• —heavy validation on large repos may be slow.

Best use case

Use this for:

• —personal projects,
• —prototypes,
• —small to medium repos,
• —code review and implementation passes,
• —React / Python / Apps Script work you want accessible from the web.

How it works

1. 1.You upload a .zip of your project.
2. 2.The server extracts it into a temporary workspace.
3. 3.The multi-agent system plans, retrieves context, edits files, validates, and reviews.
4. 4.You download the modified project as a .zip.

Beginner setup

Start here if you have never done this before:

• —`SETUP_FROM_ZERO.md`
• —`STORAGE_ESTIMATES.md`
• —`ARCHITECTURE.md`

Environment variables

Copy backend/.env.example into backend/.env for local testing, or add the same values as Space secrets/environment variables.

The most important ones are:

• —LLM_PROVIDER=cloudflare
• —CLOUDFLARE_ACCOUNT_ID=...
• —CLOUDFLARE_API_TOKEN=...

Local test run before deploying

cd backend
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env
uvicorn app.main:app --reload --port 8000

In another terminal:

cd frontend
npm install
npm run dev

Docker / Hugging Face Space deployment

This repo already contains:

• —a root Dockerfile,
• —Hugging Face Space front matter in this README.md,
• —a backend that can serve the built React frontend.

So you can upload this project directly into a Docker Space.

Current free-stack limitations

• —Uploaded project storage is temporary.
• —Free hosting can sleep when unused.
• —You should download the edited zip after working.
• —Private, always-on, persistent, or GPU-backed hosting is not free.
• —Running npm install for arbitrary uploaded repos is intentionally not automated here because that is risky on a shared free host.

Included tools

• —read_project_tree
• —read_file
• —read_files
• —write_file
• —replace_in_file
• —search_workspace
• —semantic_search_workspace
• —search_knowledge
• —read_related_files
• —inspect_dependency_graph
• —run_command
• —run_python
• —run_clasp

Good starter prompts

• —“Read this project first and propose a senior-engineer plan before changing anything.”
• —“Improve this React screen to production quality and keep the existing style patterns.”
• —“Refactor this Python module for cleaner boundaries and safer error handling.”
• —“Improve this Google Apps Script to reduce Sheet API calls and be safer for triggers.”
• —“Inspect related files before editing and explain every validation result honestly.”