Docs Tokens and AI

Prompt Chains

Pro

Watch the videoPrompt Chains1:34
Open on YouTube

A chain runs several prompts in sequence, feeding each step the previous step's output. One hotkey, several passes.

Chain: Polish Fix grammar ↓ Rewrite concisely ↓ Friendly tone
Result Correct, tight, and warm. One keystroke.

Seeded chains

Chain Abbreviation Steps
Polish cpolish Fix grammar → Rewrite concisely → Friendly tone
Translate & bullet ctransbullets Translate to English → Convert to bullet list
Explain then summarize cexplain Explain code → Summarize in 3 bullets

Building one

Create a snippet, set its type to Chain, then add steps. Each step points at another snippet. Two tokens matter inside a chain:

Token Means
{{prev}} The previous step's output. Empty on step one.
{{selection}} The original input, unchanged for the whole chain.

Selection prompts drop into chains without editing: inside a chain, TypeFire feeds {{selection}} the previous step's output, so a prompt written for standalone use reads correctly as a step.

Running one

Trigger a chain by abbreviation or by global shortcut. While it runs you get step-by-step progress, and you can cancel mid-run.

⚠️

Every step is a model call. A three-step chain takes roughly three times as long as one prompt, and on a metered provider costs three times as much. Apple Intelligence and Ollama are free per call, which makes them a good fit for chains.

Documentation for TypeFire 5.1.0. Updated 2026-09-15.