AI Prompt Toolkit

Chat Messages Studio

All tools

Assemble a chat transcript turn by turn and get request JSON for OpenAI or Anthropic.

How it works

Four simple steps

1

Add chat roles

Build system, user, assistant, and tool messages.

2

Reorder or convert

Switch between OpenAI and Anthropic payload shapes.

3

Validate JSON

Catch broken message arrays before you call an API.

4

Copy the payload

Paste into your SDK, playground, or test fixture.

Chat Messages Studio

About Chat Messages Studio

You have a working OpenAI conversation in a test fixture and now need the same exchange for Claude. Hand-editing the JSON means moving the system prompt, merging turns and rewriting tool results without breaking anything. Chat Messages Studio does that conversion and shows you what changed.

Add turns with the system, user, assistant and tool buttons, then edit each one in place. Every turn can be moved up or down, duplicated or deleted. Tool turns take a tool_call_id in the name field. The View menu switches between the OpenAI messages payload, the Anthropic request, both side by side, or a line diff of the two. In the Anthropic shape the system messages are joined into the top-level system field, consecutive same-role turns are merged, and tool turns become user turns holding a tool_result block.

Validate checks for empty turns, a conversation that starts with an assistant or tool turn, a system message that is not first, several system messages, same-role runs, tool results without an id or not following an assistant turn, and a transcript ending on the assistant. One thing to watch: the Model box only changes the OpenAI payload, so replace the model id in the Anthropic output before you send it. To keep reusable instructions separate from the transcript, draft them in the Prompt Template Builder, and size the result with the character and token counter.

The status line gives a rough token estimate for the whole conversation. Everything is built in the page; no request is sent to either provider.

Keyboard shortcut

Ctrl Enter - Run tool
FAQ

Frequently asked questions

See all FAQs

Answers for this tool. For site-wide help, open the FAQ hub.

Anthropic takes the system prompt as a separate top-level field, expects user and assistant turns to alternate, and represents content as typed blocks. The converter handles those three differences so the same conversation can be sent to either API.

It becomes a user turn containing a tool_result block, and the value you typed as tool_call_id is used as the tool_use_id. If the tool message has no id, validation flags it because both APIs need it to match the result to its call.

No. It sets the model in the OpenAI payload only. The Anthropic view uses a fixed example model id and a max_tokens value of 1024, both of which you should adjust for your own request.

View more FAQs

Related tools

View all

LLM Cost Estimator

Estimate multi-model API cost from tokens or pasted prompts.

Open

RAG Chunk Simulator

Split documents into RAG chunks with overlap, variance heatmaps, and cost estimates.

Open

Prompt Eval Rubric

Score prompt variants against weighted rubrics before spending API credits.

Open

Prompt Generator

Generate AI prompts instantly.

Open

Explore 321 free tools

Everything you need as a developer, marketer or creator - in one beautiful place.

Browse all tools