What is a prompt chain?

A prompt chain is a sequence of AI prompts where each step can reference the output of previous steps. This enables complex multi-step reasoning and processing workflows.

How do variable references work?

Use {{output_1}} to reference the output of step 1, {{output_2}} for step 2, etc. These placeholders are replaced at runtime with the actual outputs.

What format is the exported chain?

The chain is exported as a JSON array of step objects, each with systemPrompt and userPrompt fields.

AI Prompt Chain Builder

Design multi-step AI prompt chains with variable references between steps.

2 steps~32 tokens total

1Step 1

~15 tokens

System Prompt

User Prompt (use {{output_N}} for previous outputs)

2Step 2

~17 tokens

System Prompt

User Prompt (use {{output_N}} for previous outputs)

Export Chain JSON

[
  {
    "step": 1,
    "systemPrompt": "You are a research assistant.",
    "userPrompt": "Summarize the following topic: {{input}}"
  },
  {
    "step": 2,
    "systemPrompt": "You are a writing coach.",
    "userPrompt": "Improve this summary for clarity:\n\n{{output_1}}"
  }
]

Related Tools

PLVAI Pipeline VisualizerNEW

Visualize AI prompt chain JSON as a vertical flowchart.

CMPAI Response ComparatorNEW

Compare AI model outputs side by side with metrics.

FMTAI Output FormatterNEW

Auto-detect and format LLM response text as JSON, Markdown, code, or plain text.

PBDAI Prompt BuilderNEW

Build structured AI prompts with role, task, context, and output format fields.

RQBAI API Request BuilderNEW

Build AI API request payloads and cURL commands for any provider.

Learn More

guide:prompt engineering basics use case:rag pipelines

FAQ

What is a prompt chain?: A prompt chain is a sequence of AI prompts where each step can reference the output of previous steps. This enables complex multi-step reasoning and processing workflows.
How do variable references work?: Use {{output_1}} to reference the output of step 1, {{output_2}} for step 2, etc. These placeholders are replaced at runtime with the actual outputs.
What format is the exported chain?: The chain is exported as a JSON array of step objects, each with systemPrompt and userPrompt fields.

Build multi-step AI prompt workflows visually. Each step has system and user prompts. Use {{output_1}}, {{output_2}} variables to reference previous step outputs. Add, remove, and reorder steps. Export the entire chain as JSON for use in your workflow automation.