Flowcmd Llm Recipes
Use when composing the LLM-calling step of a flowcmd workflow and you need a working, backend-specific snippet — Ollama, Claude CLI, llama.cpp server, Hugging Face inference API, Unsloth / vLLM / any OpenAI-compatible server, OpenAI itself, or a generic HTTP endpoint. Covers prerequisites, expected env vars, and common JSON-escaping pitfalls.
Flowcmd Install
Use when `flowcmd` is not installed on the user's machine and they want to install it, or when a flowcmd workflow can't be run because the binary is missing. Handles installation on macOS, Linux, and Windows via the official install scripts.
Flowcmd Author
Use when the user asks to write, generate, scaffold, edit, debug, or fix a flowcmd YAML workflow — any file under a .flowcmd/ directory, any file passed to `flowcmd run` or `flowcmd validate`, or any request like "write a flow", "create a flowcmd pipeline", "chain these shell commands with an LLM in the middle". Covers the full flowcmd schema, sequential and parallel steps, retry, when-conditions…