Because the alternative is leaving your work: open a chat tab, re-explain context, wait, copy, switch back, paste, fix formatting. OpenCues answers where the question already is.
The concrete wins
- Zero context switching. The request lives inside the text; the answer lands where you put the
_. Nothing to navigate back from. - Fast enough to keep up. Word cues in 200-500ms, most fills sub-second on fast providers, static tips instantly.
- Nothing is ever forced on you. Suggestions arrive dimmed and ignorable; every substitution cycles back to exactly what you typed.
- Your model, your keys. Model agnostic with no middleman service, including fully local via Ollama.
- Extensible in minutes. A new capability is one markdown file that hot-reloads, shareable with your team via a repo's
.cues/folder. - A safety boundary you can reason about. Model output is draft text only; the system never acts on your behalf.
Related: What can OpenCues do? · vs chat panels · How it stays fast · How good is OpenCues? · Is OpenCues free? · GitHub repo