Feature suggestions

Add a list of questions already asked in the chat to make it easier to navigate

To the chatbox development and product team: Subject: Important feedback on the “Compress Conversations” feature - Request to restore the v1.18 version of the logic I'm writing to give feedback on a major and unfortunate change to the core “shrink conversations” feature between v1.16.1 and v1.19.0. In v1.16.1 (2025-09-17) , you introduced the “Compress conversations” feature. Its initial powerful and effective implementation was to allow users to generate an intelligent summary of the current conversation and then use that digest as a system tip/instruction to launch a brand new chat session . This allows the logic, intent, and context of previous conversations to be seamlessly carried over to a new stage, effectively enabling long-term, project-like collaboration with AI by overcoming the limitations of the context window. However, in v1.19.0 (2026-02-11) , this functionality was silently and completely changed. The digest is now inserted as an ordinary user message into the same chat thread. This change completely reduces the usefulness of this feature: Loss of core functionality: It no longer creates a new, manageable chat session. The compressed summary and subsequent conversations are buried in the growing, increasingly lengthy and cluttered history of existing chats, making organization impossible. Workflow broken: The original purpose of the feature — to provide a clear, authoritative “brief” for the new discussion phase — has disappeared. AI no longer sees summaries as high-priority instructions (system prompts), but just another piece of context. Poor user experience (automatically triggered) : The new “automatic compression at around 60% of the context length” mechanism is disruptive. It grossly interrupts deep, smooth conversations, simply to save the minuscule number of tokens, which prioritizes system management and ignores the user's intention and creative thought flow. For users working on complex, lengthy reasoning and projects (a key advantage of advanced AI), the v1.18 era's “compressed conversation” isn't a “token saver” — it's an essential workflow engine. Strongly requested: I strongly urge the development and product teams to revert the “compression conversation” feature to the pre-v1.19 logic, that is, to the state seen in version v1.18.x. Please restore its ability to generate a summary and use this as a system prompt to launch a new, standalone chat session. Ideally, make this the default behavior again, or at least provide it as a clear and unequivocal user option. The current implementation solves a (minor) problem that doesn't exist for advanced users, but it destroys a critical and unique ability. Thank you for your attention.

To the Chatbox Development & Product Team, Subject: Critical Feedback on the "Compress Conversation" Feature - Request to Restore v1.18 Logic I am writing to provide feedback regarding a significant and regrettable change made to the core "Compress Conversation" function between versions v1.16.1 and v1.19.0. In v1.16.1 (2025-09-17) , you introduced the "Compress Conversation" feature. Its original and powerful implementation allowed users to generate an intelligent summary of the current dialogue, which was then used as a System Prompt / Instruction to launch a brand new chat session . This enabled seamless inheritance of the previous conversation's logic, intent, and context into a new phase, effectively allowing for long-term, project-style collaboration with the AI by overcoming context window limitations. However, in v1.19.0 (2026-02-11) , this feature was silently and fundamentally altered. The summary is now inserted as a regular User Message within the same chat thread. This change has completely neutered the feature's utility: Loss of Core Functionality: It no longer creates a new, manageable chat session. The compressed summary and subsequent dialogue are buried within the existing, increasingly long and unwieldy chat history, making organization impossible. Broken Workflow: The feature's original purpose—to provide a clean, authoritative "brief" for a new phase of discussion—is gone. The AI no longer treats the summary as a high-priority directive (system prompt) but merely as another piece of context. Poor User Experience (Automatic Trigger): The new "automatic compression" at ~60% context length is disruptive. It brutally interrupts deep, flowing conversations for marginal token savings, prioritizing system management over user intent and creative flow. For users engaged in complex, long-form reasoning and projects (a key strength of advanced AI), the v1.18-era "Compress Conversation" was not a "token saver"—it was an essential workflow engine. Strong Request: I strongly urge the development and product team to revert the "Compress Conversation" feature to its pre-v1.19 logic, as seen in v1.18.x. Please restore its ability to generate a summary and use it as a System Prompt to launch a new, distinct chat session. Ideally, make this the default behavior again, or at least provide it as a clear user option. The current implementation solves a non-existent problem (for power users) while destroying a critical, unique capability. Thank you for your consideration.

Optimize image input: when I input a very long screenshot, the software compresses it, resulting in blurriness that prevents OCR recognition.

Suggest adding browser extension functionality to support summarizing webpage content while browsing.

There is a memory function, so AI can memorize key chat content based on the chat, and the window is lost every time without compressing it. The chat content in the previous window is lost in the next window. This includes recording the user's favorite way of writing. Most AI software such as Tavo, Lovemo, Cherry Studio, and RikkaHub have memory banks. It's best to have a memory, preset messages, and keep historical versions without overwriting them after modifying prompts. You can switch left and right, configure the mapping model yourself, chat suggestion functions, and not swallow records.

The knowledge base supports export and backup, allowing it to be imported onto another device or share with others.