1.8.9

Important Changes

• REST API URL uses /chat/completions endpoints. Make sure you add the correct urls that lead to /chat/completions. For example, inserting https://openrouter.ai/api/v1 will fetch https://openrouter.ai/api/v1/chat/completions. If you are using the default LM Studio url, you can insert the REST API URL as http://localhost:1234/v1.

Features

• Replaced marked for Obsidian Markdown rendering. This allows BMO Chatbot to render links, images, and other rich text elements (e.g. Obsidian Dataview, iframe, ...).
• Added Anthropic models: claude-3-sonnet-20240229 and claude-3-sonnet-20240229.

Refactor

• The model list will fetch for models every time the user open BMO Settings.
• Anthropic API's Text Completion -> Anthropic API's Messages

Fixes

• 'Prompt Select Generate' and Title Rename are updated for Anthropic Models.
• Anthropic's user response no longer removes the first word.

1.8.8

Feat

• Customizable chatbox

Improvements

• Changed temperature slider to a textfield.
  • More precise values for temperature.
• Better error handling (e.g. API connections errors will display as a bot message).
• Better command response (e.g. /prompt will display a bot message if prompt path is not set).

Fixes

• Append button will no longer go to the first file that is opened when restarting Obsidian.
• "handle responses.json when it is an array" by @keriati in (#51)

1.8.7

Add

• Bot Message now contains an Edit button.
• Google Gemini Pro's API Key.
• Mistral AI's API Key.
• Anthropic's API Key.

Changes

• The 'Prompt Select Generate' hotkey default is now CMD+Shift+=

Fixes

• Clear reference current note each time before getting the new reference note.
• Temperature max range is now set to 2.
• /save and /append now responds with a bot message when the commands are executed. This also fixes the issue where users cannot continue a conversation after /save or /append.
• Set max_tokens default to 4096 for REST API URLs, Mistral AI, and Google Gemini Pro.

1.8.6

Add

• Ollama: keep_alive parameter
• Editor Settings section
  • Prompt Select Generate System textfield

Changes

• New loading animation.
• Refactor 'Reference Current Note' to better detect active file.

1.8.5

Fixes

• Ollama: Cleared default seed parameter to avoid repetitive response

1.8.4

Added

• Ollama parameters
• Better light theme support
• Setting tabs has toggle features

Changes

• Removed gpt-4-1106-preview and replaced with gpt-4-turbo-preview

Fixes

• Persistent 'undefined' response via prompt function

1.8.3

Added

• Edit and regen button for all user messages (Excluding commands).

1.8.2

Fix

• Pulling models

1.8.1

Added

• Allow header option to display chabot name and model name.

Refactor

• Added a new notice generating... for 'Prompt Select Generate' command.
• LOCALAI REST API URL is now OPENAI REST API URL. This should support other LLM providers with OpenAI's endpoints such as LM Studio.

Fixes

• Ollama default url is set to empty to prevent repeating connection error.
• /prompt will send a new notice if folder path is not set.
• Reference current notes clears before each response.

1.8.0

Features

• Append model's response button
• Regenerate model's response button
• Generate new title command
• 'Prompt Select Generate' command
  • Ability to generate response in editor by creating a prompt, selecting the prompt, and running the command.
• Prompt option

Fixes

• Openai-based url will list the proper models in dropdown.

Deprecated

• /list
  • Use /model or /models to display model list.
  • Use /prompt or /prompts to display prompt list.