TypingMind

Similar to how we can now pin AI characters to the main screen, it would be equally useful to be able to do the same with prompts from the prompt library. Select a character in one click, select a prompt in one click. Bingo.

OpenRouter is supporting prompt caching for Claude models. Pls bring this feature to TypingMind 🙏

The Prompt Library is a crucial part of TypingMind, but with a larger number of prompts, it can quickly become disorganized. To improve usability, I would like to request the following enhancements: Adjustable Sorting Options: The ability to sort prompts based on various criteria such as alphanumeric order (both ascending and descending), most recently used, and similar options. This is especially important for me because I often create sequentially ordered prompts under specific tags, which need to be displayed in a particular order. Alphanumeric sorting would be extremely beneficial in this context. Expanded Tag System with Exclusion: I would also like the option to exclude certain tags from displaying prompts, in addition to the current tag-based filtering system. This would allow for more precise control over which prompts are visible at any given time.

In your great prompt library, add "Prompt Shelf" categories to the library + add a management feature to allow/prevent access to "prompt shelf" by users. This would be useful to prevent users from accessing prompts that are on shelfs they are not supposed to have access to (internal prompts versus public prompts, admin prompts versus user prompts, etc...). Also good to not overwhelm users with a prompt library they have no clue about.

Often when writing very long prompts I have pieces that I re-use between different prompts, and a prompt involves pulling in several of those and adding specific data or requirements. If we had a library of saved snippets that we could manage and just reference in a prompt, it would save a ton of time and make it easier to do more complex promting. Almost a way to modularize prompt building. Example: When using an LLM to generate board game turn summaries I would have a long prompt that included: Player list, description of the game rules and how to interpret moves, move list, and final sentence asking for the summary. The game rules description was always the same, and the player list only changed per game. Also, if I had a place to store labelled blocks of text, I probably would have put the turn lists into a snippet (named Turn 5 move list) so that I could try different generations with the data without having to copy and paste.

Every time I type @ I get that popup which can get very annoying. @ is used a lot in Razor/Blazor. Please let me disable that feature.

Sometimes I'd like to switch models during a prompt chain/queue. For example, I may want to create a prompt to run the first part of the prompt on GPT-4o and then apply the results to the second part of the prompt and run that on Gemini 2.0.

Hi Typingmind Loving #Typingmind thusfar, please keep up your great work. I am new to Typingmind and could not see any FAQ sections for adding something like this I am trying to see see if there is a way to run commands commands by selecting different prompts/patterns based on this project https://github.com/danielmiessler/fabric/discussions/776 if I have Fabric running on a machine somewhere. I have also posted a question on their Github page, to see if this something that can be done and pursued as I am not a developer. Ssummary of the Fabric project: #Fabric is an open-source framework for augmenting humans using AI. It offers a modular architecture for solving specific problems by leveraging a crowdsourced set of AI prompts that can be applied anywhere. The framework is designed to integrate AI into daily tasks and workflows, making it easier for users to apply AI solutions to various problems. Key features of Fabric include: • Modular framework: Fabric consists of a collection of AI prompts that can be used to solve specific problems. These prompts can be combined and customized to fit various use cases. • Crowdsourced AI prompts: The framework relies on a community-driven approach, where users can contribute and edit AI prompts to expand the framework’s capabilities. • Direct pattern referencing: Fabric allows direct referencing and editing of individual patterns, enabling users to cleanly integrate AI functionality into their applications or workflows. • Ability to directly reference and edit individual Patterns (Markdown-based format) • Custom Patterns for local use, stored in ~/.config/custom-fabric-patterns • Integration with Mill (a personal AI assistant) for automated updates and ecosystem management Would love to hear back if this something user of Typing-mind can via its UI ? Much appreciated Regards

Someone fucked up 1 line of code as it works for "---" and more dashes and yes, that includes 100+. The goal of the change was to move it from "---" to "----" as the former was used by Claude to separate parts of responses. Now it is worse than ever.