-
v.1.5.4 Stable
released this
2026-06-29 23:13:42 +00:00 | 0 commits to main since this release- Fixed an issue where the final response generation would return a tool use instead of actual text after the maximum tool use limit was exhausted.
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
v1.5.3 Stable
released this
2026-06-20 21:35:26 +00:00 | 1 commits to main since this release- Fixed an issue where a restart had to be issued for the bot to start listening to events in rooms joined after the bot was started.
- Fixed an issue where the typing indicator blocked the asynchronous execution of event handling.
- Fixed an issue where the typing indicator would disappear after the first message was sent when multiple messages were generated for the same room.
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
v1.5.2 Stable
released this
2026-06-19 17:39:04 +00:00 | 5 commits to main since this release- Fixed an issue where mentions were incorrectly logged if the bot and another user were tagged in the same message
- Refactored message reply retrieval logic to support multiple reply_to message fetches
- Added support for fetching up to 5 prior messages in the message reply chain to use it as context.
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
v1.5.1 Stable
released this
2026-05-26 21:06:53 +00:00 | 12 commits to main since this release- Fixed an issue where empty LLM response message caused by failed tool use or content moderation trigger would crash the program
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
v1.5.0 Stable
released this
2026-05-26 11:21:15 +00:00 | 14 commits to main since this releaseChanges
- Added support for handling stickers as image events
- Added image handler for when a user tags the bot while uploading an image
- Added a typing indicator decorator to mark event handlers that require sending typing indicator in chat
- Introduced a dedicated LLM response service as a middle layer for response generation and centralized error handling
- Added prompt tagging and XML tags across message history, message context, and reply prompts to improve LLM coherence, accuracy, and API error interpretation
- Added support for adding LLM-specific built-in tools to the tool registry
- Added web search and website scraping tools
- Added support for multiple tool calls (up to 10)
- Made debug level app-exclusive
- Removed duplicated error handling in event handlers in favor of the new response service
- Migrated to async OpenAI client
- Added support for both async and sync tool execution
- Added missing mxc_url check for image message events
- Added bad event checking for reply message fetching
Bug Fixes
- Fixed missing user_id context that caused the "who mentioned me last" Matrix tool to fail
- Fixed a logic bug where reply mention logging had the mentioning user and mention target swapped
- Fixed an issue where the typing indicator disappeared long before the reply generation ended
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
v1.4.0 Stable
released this
2026-05-18 16:07:12 +00:00 | 46 commits to main since this release- Added functions callable by llms during content generation
- networking: ping_host, traceroute
- time: get_date_time
- matrix: who_mentioned (to find out who mentioned a specific user, when, and in which message)
- Added mention event log service to keep track of mentions in chat rooms
- Added support for interacting with image input (if the llm model supports vision, else processes prompt without image attachment)
- Major refactoring done to project architecture.
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
- Added functions callable by llms during content generation
-
v1.4.0-pre2 Pre-release
released this
2026-05-15 22:19:11 +00:00 | 57 commits to main since this release- Added functions callable by llms during content generation
- networking: ping_host, traceroute
- time: get_date_time
- matrix: who_mentioned (to find out who mentioned a specific user, when, and in which message)
- Added mention event log service to keep track of mentions in chat rooms
- Moved files into their respective domain specific dirs: service, model, tool, etc.
- Added support for interacting with image input (if the llm model supports vision, else processes prompt without image attachment)
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
- Added functions callable by llms during content generation
-
v1.2.0 Stable
released this
2026-05-08 23:58:15 +00:00 | 90 commits to main since this release- Added support for expanding context if the original message is a reply to another
- Added support for continued conversation by replying to the bot's previous message
- Added a typing indicator while a response is being generated
- Removed
matrix_display_namefromconfig/settings.yaml - Added
matrix_join_on_inviteoption toconfig/settings.yaml
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads
-
v1.3.0 Stable
released this
2026-05-08 23:55:37 +00:00 | 91 commits to main since this release- Completely overhauled settings.yaml layout and config loading logic
- Added optional config options to specify model temperature, top_p, reasoning_effort, max_tokens and extra body
- Fixed an issue where the bot would refer to the wrong person if a reply was included in the original message
Downloads
-
Source code (ZIP)
1 download
-
Source code (TAR.GZ)
0 downloads
-
v1.1.0 Stable
released this
2026-05-08 20:48:56 +00:00 | 96 commits to main since this release- Added support for all LLM providers who expose OpenAI compatible endpoints
- Renamed config options in settings.yaml to make them more generic
- Added api_base_url config option required for specifying the OpenAI compatible endpoint base url
Downloads
-
Source code (ZIP)
0 downloads
-
Source code (TAR.GZ)
0 downloads