Skip to content

v1.1.0
Compare
Choose a tag to compare
@afadingthought afadingthought released this 08 Apr 21:12
· 47 commits to main since this release
v1.1.0
4d71b8c
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

ANTHROPIC’S API SUPPORT.

  • NOTE: The Frequency Penalty and Presence Penalty parameters are not available through the Claude’s API. Any values you insert for those parameters in your Kiki workflow will be ignored. Other than that, it works the same as OpenRouter and OpenAI models.

WHISPER AI INTEGRATION

  • If you want to “Record & Transcribe,” Kiki requires THIS SHORTCUT. The shortcut should not be renamed for it to work. (If you know of a better & free alternative that would work with Alfred to record -> save audio -> output saved path using AppleScript or a CLI, I’m open to suggestions).
  • There’s two new available Hotkey triggers. One of them will trigger an audio recording for it to be transcribed and immediately pasted in the frontmost app. The second one will trigger a recording and will ask the user for the following action.
  • The Kiki Options (“ko”) keyword now also offers the option or recording & transcribing. Press SHIFT to send the transcription through a preset right after, CMD to use it as a chat prompt, or OPT to paste transcription in frontmost app.
  • There’s new parameters that can be used for the “All-in-One” trigger to work with Whisper AI transcriptions. READ DOCUMENTATION.
  • Optionally, you can now trigger Whisper AI Transcription on audio files. It works on mp3, wav, m4a, webm, and mp4 files. Keep in mind there’s a 25 mb limit when using OpenAI’s Whisper API.
  • New “Whisper Prompt” field can be set during the configuration of the workflow.

“CUSTOM MODEL” OPTION.

  • When initiating a chat or when triggering a preset, hold FN + SHIFT modifiers and Kiki will allow the user to select any model from the user’s “models.json” file in the users preset folder. This also now allows you to change model mid-conversation (by closing the dialog, and resuming the conversation from the “ko” keyword or context file while holding down the modifiers). You do have to rename your presets folder or it to be discoverable and have configured some models for this to work. Read the INTERMEDIATE section of the guide if you don't know how.
  • These modifiers can also be used externally with the “All-in-One” trigger.
  • The FN + SHIFT modifiers work in combination with other modifiers when user is about to initiate a chat or run a preset.

OTHER IMPROVEMENTS

  • The AI responses in the AppleScript dialog are now truncated at 1800 characters to address a limitation where long responses could cause the dialog buttons to become hidden. This is a temporary workaround to display large amounts of text, while still allowing the user to access the full conversation by copying the response. I'd love to find some time to figure out how to integrate Kiki with Alfred's new text view but I can't promise anything on this yet.
  • I added a "copy" option to the AppleScript dialogs. As mentioned above, full response—even when it appears truncated—will be copied.
  • Basic file actions now also work on txt or markdown files.
Assets 3
Loading