AI Code Roaster
中文文档
A VSCode extension that roasts your code with AI-powered humor and wit.
AI Code Roaster analyzes your code files and generates entertaining feedback through 6 different AI personalities. Whether you want a brutal reality check or an ego boost, there's a role for you.
Features
- Real-time streaming responses from compatible LLM APIs
- 6 AI personalities with distinct styles (3 Chinese + 3 English)
- Custom Role System - Create, edit, and delete your own AI personalities
- Smart caching with intelligent change detection
- Auto-refresh when sidebar is visible
- Secure API key storage using VSCode secrets API
- 100+ supported file types including JavaScript, TypeScript, Python, Go, Rust, Java, and more
Quick Start
Get started in under 2 minutes:
- Install the extension from VSCode Marketplace
- Open the sidebar: Explorer → "AI Code Roaster" panel
- Configure API: Click the gear icon → "Configure API"
- Enter your credentials:
- Base URL (e.g.,
https://openrouter.ai/api/v1)
- Model name (e.g.,
nvidia/nemotron-3-nano-30b-a3b:free)
- API Key
- Open any code file and enjoy the roast!
Free APIs
Configuration
API Settings
Configure your LLM API credentials via the command palette:
- Press
Ctrl+Shift+P (Windows/Linux) or Cmd+Shift+P (macOS)
- Run
AI Code Roaster: Configure API
Required settings:
| Setting |
Description |
Example |
| Base URL |
Your LLM API endpoint (OpenAI-compatible) |
https://openrouter.ai/api/v1 |
| Model |
The model to use for generation |
nvidia/nemotron-3-nano-30b-a3b:free |
| API Key |
Your API key (stored securely) |
sk-... |
URL Requirements:
- Must use HTTPS (for production) OR
http://localhost (for local development)
Commands
| Command |
Description |
aiCodeRoaster.configureApi |
Configure API settings |
aiCodeRoaster.deleteApiConfig |
Delete all configuration |
aiCodeRoaster.switchRole |
Switch AI personality |
aiCodeRoaster.createCustomRole |
Create custom role |
aiCodeRoaster.editCustomRole |
Edit custom role |
aiCodeRoaster.deleteCustomRole |
Delete custom role |
AI Personalities
Choose from 6 different AI personalities with unique styles:
Chinese Roles
| Role |
Name |
Style |
Description |
cn-roaster |
🔥 锐评家 |
Concise, sarcastic |
Sharp-witted criticism with Chinese humor |
cn-praiser |
✨ 夸夸家 |
Enthusiastic |
Extravagant praise that goes hard |
cn-reviewer |
🔍 代码审查官 |
Professional |
Rigorous, professional code review |
English Roles
| Role |
Name |
Style |
Description |
en-roaster |
🔥 Code Critic |
Snarky, punchy |
Witty criticism with sarcasm |
en-praiser |
✨ Code Hype-man |
Energetic |
Hype up your code with enthusiasm |
en-reviewer |
🔍 Code Reviewer |
Professional |
Rigorous, professional code review |
Switch roles: Click the account icon in the sidebar header or use AI Code Roaster: Switch Role.
Custom Role System
Create your own AI personalities with the Custom Role System:
Creating Custom Roles
- Open the Command Palette (
Ctrl+Shift+P or Cmd+Shift+P)
- Run
AI Code Roaster: Create Custom Role
- Provide the following information:
- Role ID: A unique identifier (e.g.,
my-custom-roaster)
- Name: Display name (e.g.,
🎨 My Custom Role)
- Description: Short description of the role's style
- Header: Title displayed before the response
- System Prompt: The AI system prompt that defines the role's behavior
Managing Custom Roles
- Edit: Use
AI Code Roaster: Edit Custom Role to modify existing custom roles
- Delete: Use
AI Code Roaster: Delete Custom Role to remove custom roles
- Custom roles are stored in
globalStorageUri/customRoles.json
Role Validation
- Role IDs must contain only letters, numbers, hyphens, and underscores
- Custom role IDs cannot conflict with predefined roles
- All fields (name, description, header, systemPrompt) are required
Technical Details
File Validation
The extension validates files in a tiered pipeline (optimized for performance):
- File type check - Validates against supported extensions
- File size check - Maximum 100KB
- Empty content check - Skips empty files
Supported File Types
JavaScript/TypeScript: .js .ts .jsx .tsx .mjs .cjs .mts .cts
Frontend Frameworks: .vue .svelte .astro
Backend Languages: .py .rb .go .rs .java .kt .kts .swift
C/C++/C#: .c .cpp .cc .cxx .h .hpp .hxx .cs
PHP: .php .phtml
JVM Languages: .scala .sc .clj .cljs .groovy
Functional: .hs .lhs .ml .mli .re .rei .fs .fsi .fsx
Elixir/Erlang: .ex .exs .erl .hrl
Scripting: .lua .r .rmd .pl .pm .t .ps1 .psm1 .psd1
Shell Scripts: .sh .bash .zsh .fish .csh .tcsh .cmd .bat
Dart/Flutter: .dart
Other Compiled: .nim .cr .jl .v .zig .mo .lok
WebAssembly: .wasm .wat
Styling: .css .scss .sass .less .styl .stylus
Markup & Templates: .html .htm .xml .svg .xhtml .md .markdown .rst .adoc
Data Formats: .json .yaml .yml .toml .ini .cfg .conf .graphql .gql
Config Files: .env .dockerfile .makefile .mk .cmake
Protobuf & IDL: .proto .thrift .avsc
Template Engines: .erb .ejs .hbs .mustache .twig .jinja .jinja2
SQL: .sql .psql
MATLAB/Mathematica: .m .wl .nb
Misc: .ahk .vim .el .lisp
Smart Caching
The extension intelligently decides when to re-roast:
- Role change → Immediate re-roast
- >10 lines changed → Re-roast
- >60 seconds with changes → Re-roast
- Otherwise → Show cached response
This minimizes API calls while keeping feedback fresh.
Auto-Refresh
The sidebar automatically refreshes every 10 seconds when visible, ensuring you always see the latest feedback.
Development
Prerequisites
- Node.js
- VSCode Engine ^1.74.0
Build Commands
npm install # Install dependencies
npm run compile # Compile TypeScript
npm run watch # Watch mode for development
npm run lint # Run ESLint
npm run test # Run tests
Project Structure
ai-code-roaster/
├── src/
│ ├── extension.ts # Main entry point, ViewProvider
│ ├── apiClient.ts # LLM API client (streaming)
│ ├── configurationManager.ts # Secure config storage
│ ├── roastHistory.ts # Smart caching with diff
│ ├── webviewContent.ts # HTML generation
│ ├── roleManager.ts # Role management (predefined + custom)
│ └── constants.ts # Roles, file types, constants
└── package.json
License
GPL-3.0-only
For more details, see the 中文文档.
