* Agentic is a lightweight agent framework, reference implementation of the agent protocol, and an agent runtime built on Ray
* Supports teams of cooperating agents, Human-in-the-loop, easy definition and use of tools, production-tested tools
* Can be used for full browser automation, including authenticated sessions
* Example applications include daily podcast creation and meeting bot agents
* Ambrogio is an intelligent dev agent that helps tackle technical debt in your codebase.
* It systematically enhances code quality and maintainability through automated analysis and improvements.
* Starting with docstring improvements, Ambrogio uses advanced language models to understand your code and generate meaningful documentation.
* Features:
* Smart Docstring Generation: Automatically adds clear, comprehensive docstrings to classes and methods
* Multi-Provider Support: Works with various LLM providers through LiteLLM integration
* Unit Test Generation (Beta): Analyzes code coverage and generates missing unit tests
* Pre-PR test runs to prevent regressions
* Improve existing docstrings in modified methods
* Documentation generation for easier understanding
* Type safety refactoring
* Spaghetti code cleanup
* Code formatting and best practices enforcement
Vulture is a tool that analyzes Python code to identify unused imports, variables, functions, and other code elements.
**Features**
1. **Unused import detection**: Finds unused imports in the analyzed code.
2. **Unused variable detection**: Detects unused variables, including global and local ones.
3. **Unused function detection**: Identifies unused functions.
4. **Unreachable code detection**: Finds unreachable code, such as code after `return`, `break`, `continue`, or `raise` statements.
**How it works**
Vulture uses the Abstract Syntax Tree (AST) module to build a tree representation of the analyzed code. It then traverses this tree and records the names of defined and used objects. The tool reports the objects that have been defined but not used, which are considered dead code.
* Open-source TypeScript starter for full-stack web development on Cloudflare
- # Building on Strong Foundations
* Core:
+ Vite + React + React Router v7
+ Cloudflare Pages (hosting) + D1 (database)
+ TypeScript + TailwindCSS + Drizzle ORM
* Tooling:
+ GitHub Actions (CI/CD)
+ Prettier + ESLint (code quality)
+ Vitest + Playwright (testing)
+ Storybook + MDX (docs & components)
- # Essential npm Scripts
* `npm run setup`
- Performs `npm install`, setups basic env vars, database and playwright browser drivers
* `npm run dev`
- Boots development server
* `npm run build`
- Builds the app
* `npm start`
- Runs built app; this is how your app will be run on production
* `npm test`
- Executes unit tests
* `npm run test:types`
- TypeScript types check
* `npm run test:e2e:ui`
- Executes browser tests in UI mode, perfect for development
* `npm run lint`
- ESLint check
* `npm run format`
- Prettier check
* `npm run storybook`
- Boots Storybook
* `npm run doctor`
- Runs all basic sanity checks: format, lint, types check and unit tests
# Autogrep Tool
* Autogrep is a tool for automatically generating and filtering Semgrep rules from vulnerability patches.
* It addresses the need for high-quality security rules after Sem Gregg's official rules are no longer available under permissive licenses.
* Autogrep leverages Large Language Models (LLMs) to automate rule generation and validation.
* Easily transcribe YouTube videos, MP4, and MP3 files to text using Whisper on Google Colab.
* No local setup required — runs entirely on Google Colab
* Supports YouTube URLs, MP4, and MP3 files
* Outputs transcribed text in .txt format
* **ParallelChunker**: A class used to split a large codebase into equal chunks suitable for Large Language Models (LLMs).
* **Code Chunking**: The process of dividing a large codebase into smaller, more manageable chunks.
* **LLM Context Windows**: The size of the context window for LLMs, which determines how much code is processed at once.