Add LLM completions to any text file without opening a chat window
Generate inline code snippets by writing @piqo instructions in source files
Run a headless watcher over a project directory to auto-fill prompt markers
Keep research notes that auto-expand each time you save the file
Requires the Pi runtime installed first, then one npm-loaded command to start the watcher.
Piqo is an extension for a tool called Pi, which is a way to work with AI language models from your computer. Instead of opening a chat window and talking to the AI there, Piqo lets you write your question or instruction directly inside any text file. When you save the file, the AI reads what you wrote, generates an answer, and pastes that answer back into the same file. The prompt line you wrote gets removed in the process, so the file ends up holding only the generated content. The trigger is a small marker. Anywhere in a file you can write @piqo followed by an instruction, for example @piqo add a function to parse CSV files. When you save, Piqo notices the change, sends the request to the AI, and rewrites the file with the answer. This works in code files, in Markdown notes, and in config files like YAML. The author suggests it for keeping notes, writing tasks, and doing quick research. To use it, you install Pi first, then run a single command that points Piqo at one or more folders you want it to watch. You can load Piqo straight from the npm package registry, or clone the repository and reference it locally. There is also a headless mode that runs without the text interface. Some practical details: file changes are debounced at half a second so the same edit is not processed twice, hidden folders and node_modules and .git are skipped, and only common text file types are read. The repository has 20 stars and is written in TypeScript.
Generated 2026-05-22 · Model: sonnet-4-6 · Verify against the repo before relying on details.