llm-cli

The llm-cli skill enables users to process textual and multimedia content through multiple large language model providers using the `llm` command-line interface. It addresses the need for a unified, provider-agnostic workflow by allowing users to interact with OpenAI, Anthropic, Google Gemini, and local Ollama models from a single interface. Whether running one-off prompts or maintaining interactive conversations, the skill streamlines model access, input handling, and configuration management. It automatically detects available providers through environment variables and guides model selection when needed, reducing setup friction and simplifying multi-provider usage.

Core capabilities include provider detection via API-related environment variables, flexible model selection using both provider and model aliases, and persistent configuration stored in `~/.claude/llm-skill-config.json`. The skill supports multiple input types, including inline text, piped stdin, and file paths such as `.txt`, `.md`, `.json`, `.pdf`, and image files. It offers both non-interactive execution for single commands and an interactive mode for conversational workflows. When model selection is ambiguous, it presents an interactive menu, and it remembers the last used model to improve efficiency in subsequent sessions.

This skill is well-suited for developers, technical users, and automation workflows that rely on command-line tools. Common use cases include summarizing documents, analyzing files, processing structured data, conducting reasoning tasks, and chatting with remote or local LLM instances. It is especially useful in environments where users need to switch between providers, test different models, or integrate LLM processing into scripts and pipelines.