Releases: longy2k/obsidian-bmo-chatbot
Releases · longy2k/obsidian-bmo-chatbot
2.3.3
2.3.2
2.3.1
2.3.0
Important Changes
You may have to reset your settings to use the /load
command features.
What's New
/load
command to list all chat histories.- Type
/load [filename] or [value]
to load a chat history. - The formatting of a chat history must include at least two h6 headers to load into view.
- Type
Improvements
/save
command will save to the loaded chat history.
Fixes
- Fix BMO Generate temperature from int to float.
- Fix BMO Generate from creating multiple backticks when aborting.
2.2.0
What's New
- "BMO Generate" are code blocks that can generate from custom prompts and display the response in your Obsidian editor.
- SupportsMODEL
,TEMPERATURE
, andMAX_TOKENS
in the code block.
- Addedbmo_generate_background_color
to profile frontmatter.
- Addedbmo_generate_font_color
to profile frontmatter. - Added all Google Gemini models (e.g.
gemini-1.0-pro
andgemini-1.5-flash
)
Improvements
- Google Gemini models can be streamed and aborted.
- Fetch and abort model response.
- Added 'No Model' option to model dropdown.
2.1.2
2.1.1
2.1.0
Important Changes
- Ollama now requires additional setup!
- PLEASE UPDATE YOUR FRONTMATTER PROPERTIES.
allow_header
changed toenable_header
reference_current_note
changed toenable_reference_current_note
What's New
- Users can insert links such as
[[example]]
,![[example]]
, and[[example#Heading]]
to be read by the model. - Models can read language blocks such as Obsidian dataview, Obsidian Tracker, and etc.
- Models can read notes when the model responds as
![[]]
or[[]]
. - Support for Ollama Vision models. Use the
LLaVA
models from the Ollama library. - Send button. Mobile-friendly.
- Stop button for streaming models.
- Appearance option to enable/disable scrollbar in chatbox.
- Ollama
min_p
parameters. - Added 'claude-3-haiku-20240307' model from Anthropic.
- Added 'gpt-4-turbo' model from OpenAI.
- Added 'gpt-4o' model from OpenAI.
Improvements
- Mobile-support for return key will add newline instead of submission.
- Some commands are replaced with a new notice rather than a response.
- Reload button for models dropdown in General Settings.
- 'Prompt Select Generate' command will display a 'Generating...' notice until response finishes.
- Trim output of model responses.
Changes
- Removed 'gpt-4-turbo-preview' model from OpenAI in favor of 'gpt-4-turbo.'
- Removed 'gpt-3.5-turbo-1106' model from OpenAI in favor of 'gpt-3.5-turbo.'
- The textarea scrollbar is now hidden by default.
- 'REST API URL Connection' changed to 'REST API Connection' in settings.
- Username 'USER' changed to 'YOU.'
Fixes
- Fix slash ('/') issues.
- Fix
/save
command from message order by removing chat response. - Deleting default profile will reset to default settings.
2.0.0
Important Changes
CORRECTION - Semantic Versioning: Breaking changes v1.8.9 --> v2.0.0 (v1.9.0)
If you have <v2.0.0, please perform the following instructions:
- Go to
Options > Community plugins > BMO Chatbot
and uninstall the plugin. - Re-install "BMO Chatbot"
- Restart Obsidian or toggle the plugin on/off to refresh.
Or,
- Go to
Options > Community plugins
and click on the folder's icon:
- Close Obsidian completely.
- Find the
bmo-chatbot
folder and deletedata.json
. - Restart Obsidian.
What's New
- Re-implemented Prompts
Improvements
- "/prompt [PROMPT-NAME]" is now case insensitive.
Changes
- Moved "Allow Header" to the top of the Appearance settings.
- Reference current note is prepended with "Additional Notes:"
Fixes
- Semantic Versioning: Breaking changes v1.8.9 --> v2.0.0 (v1.9.0)
- Gemini Pro does not support system prompt. Resolved using @gerardsans solution.
- Fixed 'Chatbox Background Color' frontmatter. Did not update chatbot view when resetting to default.
- Help command typos
1.9.0
Breaking Changes
- BMO Chatbot's data structure has been modified which may lead to errors. Please use the 'Reset Settings' option at the bottom of the settings tab to refresh your setting's data.
If you have <v1.9.0, please perform the following instructions:
-
Go to
Options > Community plugins
and click on the folder's icon: -
Close Obsidian completely.
-
Find the
bmo-chatbot
folder and deletedata.json
. -
Restart Obsidian.
What's New
- Profiles - Create chatbots with specific knowledge, personalities, and presets.
- Each profile has their own message history.
- Switch profiles via settings or commands. Try
/p
to list profiles.
- More chatbot style customizations such as modifying the background color of the chatbot container, message container, and font colors.
- OpenRouter now have there own section under
BMO Settings > API Connections
. - A 'Reset Settings' option. Please reset your settings with this update for a better experience.
Improvements
- Command aliases
- Help (e.g. manual, commands):
/commands
,/manual
,/man
,/help
,/h
- Models:
/models
,/m
- Profiles:
/profiles
,/profiles
,/prof
,/p
- Reference Current Note:
/reference
,/ref
- Temperature:
/temperature
,/temp
- Max Tokens:
/maxtokens
- Append current message history to note:
/append
- Save message history:
/save
- Clear message history:
/clear
,/c
- Stop stream:
/stop
,/s
- Help (e.g. manual, commands):
- Models and profiles are case-insensitive when switching via command.
Changes
- Prompt Select Generate System Role default changed from "You are a helpful assistant." to "Output user request."
- Merged 'Chatbox Border Color' with 'Chatbox Background Color'.
- Scrollbar back in chatbot view.
- 'User Name' and 'Chatbot Name' can now be upper or lowercase.
- The default 'Chat History Folder Path' changed from 'BMO' to 'BMO/History'.
- Removed 'Chatbot Name' from Appearance settings. Creating and renaming chatbots are located in
Profiles > Profile Folder Path
. - Removed 'System' (System role prompt) from the settings tab. The aim is to encourage users to locate the system role prompt from the profile and create more complex prompts.
Fixes
- Fixed Anthropic Rename Title Generator
- Fix paragraph breaks when refreshing chatbot view.