[AI Dev Tools] Command-Line Assistants, LLM-Powered Linting, AI-Enhanced Development Workflows ...

Shell-ask is a command-line tool that allows users to interact with LLMs directly from their terminal, enabling natural language queries and task automation.

• Supports multiple LLMs including OpenAI, Anthropic Claude, Ollama, Google Gemini, and Groq.
• Processes natural language queries and generates appropriate shell commands or explanations.
• Allows piping of command outputs or file contents as context for queries.
• Provides options for follow-up questions, web searches, and fetching web page content.
• Offers customizable result types, including command-only output and structured data.
• Enables the creation of reusable AI commands for common tasks.
• Installation is simple via npm: `npm i -g shell-ask`.
• For example, you can generate git commit messages based on diff output or convert video files using ffmpeg with natural language instructions.

Code2Prompt is a command-line tool that generates comprehensive prompts from codebases, facilitating interactions between developers and LLMs for code analysis, documentation, and improvement tasks.

• Generates well-structured Markdown prompts that capture the entire project's essence, including a hierarchical view of the codebase structure.
• Offers customizable prompt templates using Jinja2, allowing tailored outputs for specific AI tasks.
• Implements smart token management to ensure compatibility with various LLM token limits.
• Integrates with .gitignore rules and provides flexible file handling using glob patterns for accurate project representation.
• Provides multiple output options, including clipboard copying, file saving, and console display.
• Enhances code readability by adding line numbers to source code blocks for precise referencing.
• Enables contextual understanding for LLMs, leading to more accurate suggestions, improved documentation, and efficient refactoring recommendations.
• Includes a token price estimation feature for various AI providers and models, helping developers manage costs effectively.

GPTLint uses LLMs to enforce higher-level best practices across codebases, extending traditional static analysis tools like ESLint.

• Enforces complex best practices beyond traditional AST-based approaches using LLMs.
• Uses simple markdown format for rules, with easy customization and project-specific rule creation.
• Integrates seamlessly with existing workflows, supporting the same CLI and config format as ESLint, including inline overrides and config files.
• Provides content-based caching and outputs LLM stats per run for cost and token usage tracking.
• Supports various LLM providers and local models, with extensively tested built-in rules.
• For example, you can enforce array indexing best practices or create custom rules specific to your project's needs.

llm.nvim is a Neovim plugin that enables LLM-assisted programming with a minimalist approach.

• Integrates LLM capabilities directly into Neovim, allowing users to prompt LLMs with selected text or file content up to the cursor.
• Supports multiple LLM services including Groq, OpenAI, and Anthropic, with the ability to add custom OpenAI-compatible services.
• Offers flexible configuration options, including timeout settings and service-specific parameters.
• Provides functions for creating dedicated LLM interaction files and using text objects and motions for LLM prompts.
• Installation requires setting up API keys for the desired LLM services and adding the plugin to Neovim using a package manager like lazy.nvim.
• Users can customize keybindings for various LLM interactions, such as prompting, replacing text, and using different services.

An answer engine that leverages multiple AI technologies to process user queries and return comprehensive results including sources, answers, images, videos, and follow-up questions.

• Utilizes a combination of technologies including Groq, Mistral AI's Mixtral, Langchain.JS, Brave Search, Serper API, and OpenAI for query processing and content retrieval.
• Built with Next.js and Tailwind CSS, providing a modern and responsive user interface.
• Implements RAG (Retrieval-Augmented Generation) techniques using OpenAI Embeddings and Langchain.JS for text operations.
• Offers optional features like Ollama for local inference and embeddings, and Upstash Redis for rate limiting and semantic caching.
• Includes function calling support for enhanced capabilities such as location services, shopping, stock data, and Spotify integration.

This is another project of called code2prompt, with similar functionality. Itis a CLI tool that transforms codebases into single LLM prompts, featuring source tree structures, prompt templating, and token counting.

• Generates well-formatted Markdown prompts from entire codebases, respecting .gitignore and allowing file filtering via glob patterns.
• Customizable prompt generation using Handlebars templates, with built-in templates for common use cases like code documentation and security vulnerability detection.
• Displays token count of generated prompts using various tokenizers compatible with OpenAI models.
• Includes optional features like Git diff output, automatic clipboard copying, and line numbering for source code blocks.
• Supports user-defined variables in templates, allowing for dynamic prompt customization based on user input.
• For example, you can use it to generate a Git commit message for staged files or create a GitHub Pull Request description by comparing branches.

Plock is a tool that enables users to generate text using LLMs directly from any text input field, with real-time streaming output.

• Seamless integration with any application where text can be typed. Users can write a prompt, select it, and use a keyboard shortcut to replace it with the LLM-generated output.
• Context-aware generation using clipboard content. Users can copy text as context before writing a prompt, enhancing the relevance of generated content.
• Fully local operation by default, with options to use external APIs or custom shell scripts for text generation.
• Customizable settings through a JSON file, allowing users to modify shortcuts, models, prompts, and output behavior.
• Flexible trigger system that supports chaining actions, environment variable storage, and multiple output options including streaming text, writing final text, or displaying images.

Source: CodeUpdateArena: Benchmarking Knowledge Editing on API Updates

Source: Rectifier: Code Translation with Corrector via LLMs

Source: Prompting Techniques for Secure Code Generation: A Systematic Investigation

Source: What's Wrong with Your Code Generated by Large Language Models? An Extensive Study