Recallery adds saved prompts, ordered chains and local search to AI chat sites

Tool snapshot
Best fitProductivity
In one lineA Chrome extension for organizing reusable prompts and finding messages across ChatGPT, Claude, Gemini and Grok.

Reuse the prompt without replacing the chat app

Recallery is a Chrome extension for ChatGPT, Claude, Gemini and Grok. It adds a prompt library, prompt chains, folders and full-text search to the services a user already visits.

The interface distinguishes inserting a saved prompt from running a chain. Typing // selects a reusable prompt and places it at the cursor without sending it. Typing .. selects an ordered chain, which waits for each reply before sending the next step. That distinction matters when a user wants a draft to inspect rather than a sequence that continues automatically.

Share variables and bring context deliberately

The product supports blanks and defaults in reusable prompts, with shared answers across the steps of a chain. It also describes an @@ context shortcut for adding a transcript from another conversation, including one from a different AI service.

Recallery says its library and search operate in the browser. That storage arrangement should be distinguished from where a chosen chat prompt is processed: inserting another conversation into a message can share that context with the AI service receiving the message. Review the inserted transcript before sending sensitive material across services.

Search scope depends on the plan

The public page describes free search over the latest thirty days and broader search with additional filters on Pro. Its introductory offer includes seven days of Pro without an account or card, followed by the relevant free or paid limits.

For a practical trial, save one prompt with a variable and test a short chain whose intermediate answers are easy to check. Confirm when the extension sends the next step and how search opens the original message. The value is less repetition and easier navigation; it does not independently verify the accuracy of an answer produced by the underlying assistant.