Chat Extractor

An Electron desktop tool that uses OpenAI GPT models to extract code from chat conversations and generate project file structures. Evolved over six versions from a minimal code extractor to a local-first extraction engine with metalanguage support.

Key Capabilities

• Paste chat conversations or project documentation as input.
• OpenAI GPT analyzes text and extracts/generates code.
• Extracted files organized into proper directory structure.
• Files written to disk automatically.
• v3+ adds UI features: progress bar, file preview, dark theme.
• v4 adds multi-stage pipeline: plan → generate → audit → auto-fix.
• v5 adds custom output folder selection and installer builds.
• v6 adds local-first extraction, metalanguage, content classifier, and security features.

Versions Overview

Version	Package Name	Model	Summary	Major Features	Major Issues
Wersja 1 (v1.0.0)	chat-extractor	gpt-4.1	Minimal prototype. Single-pipeline extraction.	Textarea input, LLM extraction, file writing	No progress UI, silent errors, model may not resolve
Wersja 2	chat-extractor	gpt-4.1	Chunked IPC. Same core, adds transport robustness.	Chunked IPC (5 files/batch), explicit .env resolution	Dead code (projectBuilder), fire-and-forget chunks
Wersja 3 (v3.0.0)	chat-extractor-v3	gpt-4.1-mini	Modular architecture + UI.	Dark theme, progress bar, preview modal, organized dirs, token estimation	Duplicate file overwrite, modal missing filename
Wersja 4 (v4.0.0)	ai-project-generator	gpt-4.1-mini	Multi-engine project generator.	Planner/generator/auditor/fixer pipeline, retry logic, JSON repair, atomic writes, structured logging, pause/resume, packaging	Unused refiner engine, naive JSON extraction, no rate limiting
Wersja 5 (v1.0.0)	chat-extractor	gpt-4.1	Custom output folder + installer.	Native folder picker, electron-builder NSIS installer, custom output path	API-dependent, no local extraction, no progress UI
Wersja 6 (v1.0.0)	chat-extractor	gpt-4.1	Local-first extraction engine. Complete rewrite.	Local multi-strategy parser (461 lines), metalanguage (create: syntax), content classifier, path traversal protection, atomic writes, extraction reports, Jest test suite	Heuristic parser edge cases, no dark mode, no live preview

Quickstart

General Prerequisites

• Node.js 18+
• npm
• Valid OpenAI API key (optional for v6 local mode)

Per-Version

# Wersja 1
cd "Wersja 1/production-ready" && npm install && npm start

# Wersja 2
cd "Wersja 2/production-ready" && npm install && npm start

# Wersja 3
cd "Wersja 3/production-ready" && npm install && npm start

# Wersja 4
cd "Wersja 4/production-ready/app" && npm install && npm start

# Wersja 5
cd "Wersja 5/production-ready" && npm install && npm start

# Wersja 6 (no API key needed for local mode)
cd "Wersja 6/production-ready" && npm install && npm start

Versions 1-5 require OPENAI_API_KEY — copy .env.example to .env and add your key. Version 6 works without an API key in local mode. See per-version docs for details.

Repository Structure

├── README.md               This file
├── CHANGELOG.md            Version history
├── CONTRIBUTING.md         How to add versions and write docs
├── SECURITY.md             Security guidance
├── LICENSE                 License placeholder
├── .gitignore
├── docs/
│   ├── ARCHITECTURE.md     Cross-version architecture narrative
│   ├── VERSIONS.md         Deep per-version summaries
│   ├── TESTING.md          Smoke tests + future test strategy
│   └── KNOWN_ISSUES.md     Issues per version with severity
├── Wersja 1/production-ready/   v1.0.0 — minimal prototype
│   ├── README.md, .gitignore, .env.example, LICENSE
│   ├── main.js, preload.js, renderer.js, index.html
│   ├── modules/            (extractor, parser, validator, fileWriter)
│   └── docs/               (RUN.md, FEATURES.md, LIMITATIONS.md)
├── Wersja 2/production-ready/   v2 — chunked IPC
│   ├── README.md, .gitignore, .env.example, LICENSE
│   ├── main.js, preload.js, renderer.js, index.html
│   ├── modules/            (+ projectBuilder.js)
│   └── docs/
├── Wersja 3/production-ready/   v3.0.0 — modular + UI
│   ├── README.md, .gitignore, .env.example, LICENSE
│   ├── main/, parser/, worker/, renderer/, utils/
│   └── docs/
├── Wersja 4/production-ready/   v4.0.0 — multi-engine generator
│   ├── README.md, .gitignore, .env.example, LICENSE
│   ├── app/backend/         (orchestrator, engines, stable-openai, etc.)
│   ├── app/frontend/        (UI, styles, audit viewer)
│   └── docs/
├── Wersja 5/production-ready/   v5 — custom output folder + installer
│   ├── README.md, .gitignore, .env.example, LICENSE
│   ├── main.js, preload.js, renderer.js, index.html
│   ├── modules/            (extractor, parser, validator, fileWriter)
│   └── docs/               (RUN.md, FEATURES.md, LIMITATIONS.md)
└── Wersja 6/production-ready/   v6 — local-first extraction engine
    ├── README.md, .gitignore, .env.example, LICENSE
    ├── main.js, preload.js, renderer.js, index.html, jest.config.js
    ├── modules/            (parser, extractor, classifier, metalanguage, validator, fileWriter, reporter)
    └── docs/               (RUN.md, FEATURES.md, LIMITATIONS.md, METALANGUAGE.md, ROADMAP_v3.1.md)

How Versioning Works

Each version lives in Wersja N/production-ready/ — a self-contained, GitHub-publishable directory with its own package.json, dependencies, source code, documentation, and .gitignore. Any production-ready/ folder can be copied into a new repo and pushed to GitHub as-is.

See CONTRIBUTING.md for conventions.

Documentation

Document	Description
docs/ARCHITECTURE.md	How the tool evolved across versions
docs/VERSIONS.md	Deep summary of each version
docs/TESTING.md	Smoke tests and testing strategy
docs/KNOWN_ISSUES.md	Bugs and limitations per version
SECURITY.md	Security guidance
CONTRIBUTING.md	How to contribute
CHANGELOG.md	Version changelog