v3.0.0

TL;DR

• Moved from wwjs to Baileys, making the code use way less ram and hopefully increasing stability in the long run.

What's Changed

• Baileys by @Luisotee in #312

Full Changelog: v2.13.0...v3.0.0

v2.13.0

TL;DR

• Added Docker.
• Added pulses inspired by MemGPT heartbeat, a thing that wakes the LLM periodically to give updates to the user like an event in his calendar, some route that you asked to be kept informed, etc. Generally things in the LLM core memory, conversational window and things from the tools enabled. There is also a way to ask the LLM to schedule a heartbeat, though they won't persist a reset.
• (Re) Added a queue system.
• Added a !reset lc_mem command that only deletes the conversation memory, and not the core memory.
• If you are updating you will need to update your .env variables.

What's Changed

• Docker by @Luisotee in #308
• Command to delete memory from lc conversations by @Luisotee in #309
• Pulses by @Luisotee in #310

Full Changelog: v2.12.0...v2.13.0

v2.12.0

TL;DR

• Added a core memory to the bot, this is supposed to store crucial info in a more permanent state (outside the limited memory)
• The LLM should add, delete and update those infos by itself. But you can facilitate it by telling the bot to use the tool
• When updating, you will need to do a pnpm build which shouldn't reset your database anymore.
• This is the first step towards a bunch of features that I have been wanting to add, you can check those in the projects tab

What's Changed

• GitHub models support by @Luisotee in #306
• Initial implementation of core memory by @Luisotee in #307

Full Changelog: v2.11.0...v2.12.0

v2.11.0

TL;DR

• Added support for direct integration with Google, Anthropic, and Groq APIs, leveraging Langchain's tool-calling agent to boost performance.
• Currently all Groq and Gemini models in the change command are free.
• Groq also offers Whisper for free, which has been integrated as well.
• You will need to do a pnpm i

What's Changed

• Some updates and gemini tool calling support by @Luisotee in #302
• Anthropic tool calling by @Luisotee in #303
• Groq tool calling by @Luisotee in #304
• Groq whisper by @Luisotee in #305

Full Changelog: v2.10.0...v2.11.0

v2.10.0

TL;DR

• Added a Weather Tool to Langchain
• Updated LLMs in change command

What's Changed

• Wwjs update and gpt4o by @Luisotee in #298
• Weather tool by @Luisotee in #301

Full Changelog: v2.9.0...v2.10.0

v2.9.0

TL;DR

• Added new Google Routes Tool https://js.langchain.com/v0.2/docs/integrations/tools/google_routes

What's Changed

• Langchain update + Google Routes Tool by @Luisotee in #297

Full Changelog: v2.8.0...v2.9.0

v2.8.0

TL;DR

• Changed langchain agent to tool calling in OpenAI models.
• This massively improves performance, eliminating the need to delete memory with every message and making it much more reliable. I highly recommend using an OpenAI model for now.
• In the next few days, support for tool calling will also come for Anthropic and Google models.

What's Changed

• Updated langchain to 0.1.36 by @Luisotee in #291
• Tool calling agent for OpenAI models by @Luisotee in #292

Full Changelog: v2.7.0...v2.8.0

v2.7.0

TL;DR

• Changed langchain agent to structured chat which should be better than reAct
• Tweaks in the prompt for both bing and langchain
• New .env variables DEFAULT_MODEL and MODEL_TEMPERATURE. You will need to modify your .env with them.

What's Changed

• Changed agent used in langchain to structured chat by @Luisotee in #290

Full Changelog: v2.6.0...v2.7.0

v2.6.0

What's Changed

• feat: Wikipedia tool by @Luisotee in #287

Full Changelog: v2.5.0...v2.6.0

v2.5.0

TL;DR

• Added image generation with dalle into openrouter models, more info at env.example.
• Added basic calculator tool from langchain.
• Updated langchain and fixed all error messages.

What's Changed

• chore: updated langchain to 0.1.28 by @Luisotee in #284
• feat: Dalle tool and basic Calculator tool by @Luisotee in #286

Full Changelog: v2.4.1...v2.5.0