etherealengine / digital-beings Goto Github PK

• Get the Digital Being code etc you need from alextoti
• Make the bot functions work nice in a group setting by:
• Let the bot answer to being mentioned or answer to its first name or answer to its 2nd name if it has a 2nd name
• Make sure it doesn't embed so it looks natural.
• If you mention someone other than the bot, the bot won't respond.
• If you continue to talk without mentioning the bot won't respond.
• If you use the name of the bot or mention the bot the bot will respond
• After this you don't have to name the bot or mention the bot for future conversation
• If the bot is having multiple conversations it will mention the users so you know who it's talking to..
• Make the bot answer recently edited questions and delete the answer to the pre edited question... Disregard if you don't have delete or edit feature.. If it just has edit than edit the answer to reflect the newly edited question.
• For all the messages that the bot is asked not to respond to, let it passively put them to memory.
  set it up so it doesn't ever crash and log what would have caused the crash and set it up so the ai will send a direct message stating which line the error is on and the stacktrace. Alextoti set this up for discord so any questions about this should go to him
• For Kyle. Get them to add branding that indicates the AI is from the Metaintelligence Project to Harmony. Remind Shane to give the logo.
• Once it's ready for testing.. let me know so I can help check to see it's working good

We need a general command system that can forward commands directly to the python client (so if we write /say it will send that into the agent to process in it's own command language)

It should log as part of chat history as an event, not as part of conversation history.

The goal is to have the bot connect to a room in XREngine from Discord so that we can easily control it remotely without needing to restart the bot, etc.

Work with The Researcher to ensure that the bot can successfully emote or walk around successfully as a human would.

We want to pipe all of these into chat but hide them from most users, or make them something that can be subscribed to.

A good way to do this may be something like /getAllEvents or something in chat, so that we can use our chat handler API.

The commands are:

User Joins

User Leaves

User comes close or moves further away

User comes to intimate distance or moves further away

User is looking at me

User stops looking at me

We want audio to come in and out from webrtc to our python layer via gRPC

When the bot joins, we want to give them a list of users

Maybe we can drive this through chat to make it easy.

For example, /listAllusers in chat will recall the list

Clone of #130

https://github.com/hyunwoongko/openchat from OpenChat

Hello. I was installing DigitalBeing and had some issues. I'd make PR, but want to know what others think first, because I barely understand the project's internals.

DockerHub shows a bit more config options for Postgres than currently are used.
My approach here would be not to expect the user to configure things manually as it's told in the readme step 7, but to have a preconfigured Postgres instance instead.

This is what we have for Docker now:

  db:
    image: postgres 

Instead, we can configure docker-compose.yml with some defaults:

  db:
    restart: always
    image: postgres
    hostname: digitalbeing_db
    environment:
      POSTGRES_DB: digitalbeing
      POSTGRES_USER: digitalbeing
      POSTGRES_PASSWORD: digitalbeing_default_pw
    volumes:
      - ./init.sql:/docker-entrypoint-initdb.d/init.sql

and have a "starter" SQL file as well, instead of making the user manually do the parts of the installation process.

-- init.sql
GRANT ALL PRIVILEGES ON DATABASE digitalbeing TO digitalbeing;
CREATE TABLE chat_history(client_name text, chat_id text, message_id text, global_message_id text, sender text, content text);

& said the bot responses need to be saved on xr database so they can retrieve it when they want to . So I am now planning to work on that discussing with alex . The responses can be multiple according to & like if I say 'hi' to the agent it can reply like
'line 1 .......'
'line 2 ......'
'line 3 .....'
and also we have to map users and many things yet to come from & to work on this API .
but the basic requirement starts from this

right now all the agents reply at once . like if I say 'hi' the reply is something like 'rasa: reply...','gpt-neo: reply ..','gpt3: reply:...,we will create separate bot for each agent and calling the bots it can get responses from agents separate like this@gpt-neo hi gpt-neo replies -> hello sifan, how are you

This issue lists Renovate updates and detected dependencies. Read the Dependency Dashboard docs to learn more.

Rate-Limited

These updates are currently rate-limited. Click on a checkbox below to force their creation now.

• Update dependency nodemon to v2.0.21
• Update dependency puppeteer-extra to v3.3.6
• Update dependency puppeteer-extra-plugin-stealth to v2.11.2
• Update dependency discord.js to v14.7.1
• Update dependency eslint to v8.35.0
• Update dependency eslint-plugin-react to v7.32.2
• Update dependency node-telegram-bot-api to v0.61.0 (node-telegram-bot-api, @types/node-telegram-bot-api)
• Update dependency openai to v0.27.0
• Update dependency pg to v8.9.0
• Update dependency redis to v4.6.5
• Update dependency twilio to v3.84.1
• Update dependency twilio to v7.16.4
• Update dependency twitter-api-v2 to v1.14.2
• Update dependency typescript to v4.9.5
• Update node orb to v5.1.0
• Update dependency @types/node to v18
• Update dependency axios to v1
• Update dependency puppeteer to v19
• Update dependency semantic-release to v20
• Update dependency twilio to v4
• 🔐 Create all rate-limited PRs at once 🔐

Open

These updates have all been created already. Click a checkbox below to force a retry/rebase of any.

• Update dependency @types/recordrtc to v5.6.10
• Update dependency config to v3.3.9
• Update dependency express to v4.18.2 (express, @types/express)
• Update dependency flask to v2.2.3
• Update dependency puppeteer to v17.1.3
• Update react monorepo (@types/react, @types/react-dom)
• Update dependency @types/node to v16.18.14
• Update dependency @types/three to v0.149.0
• Update dependency transformers to v4.26.1
• Update dependency uuid to v9
• Click on this checkbox to rebase all open PRs at once

Detected dependencies

• Check this box to trigger a request for Renovate to run again on this repository

When any NLP agent is processing messages, converting something like Discord mentions (<@!\d+> format) in both messages and history to nicknames or names may help a lot compared to Discord IDs represented as strings.

The example.py version I have receives Discord messages in a format like <@!000000000000000000> test". I am not sure about the other interfaces, but they probably have similar mechanisms.

(This is more natural, and is what a human would do.)

Reasoning: Instagram DM is the platform almost all teenagers use in the West– much more popular than Discord, in fact.
Support replies, likes and other reactions

This will let us run Metaintellgince bridge and Digitla Beings bridge on one Discord bot account.

Should be implemented, but definitely needs to be tested and documented

Echo bot should do every possible action the bot can do on startup, and also check inputs.

Discord: Online -> Idle, Online -> Offline

We should send the events over to the agent

We should already have a chat history route on the API server. If we can recall that and wire it to the Digital Being to test, maybe through chat window, that could be best.

i.e. /getChatHistory or something in the chat?

When lurking: Send thread titles
When replying to a thread: Include the link description

https://github.com/hyunwoongko/openchat from OpenChat

Store conversations on the bot side. Keep logs for each user for each channel, to be able to recall and continue from last conversation.

Liam says we already have a SQLite database in the Digital Being repo that should do this.

& said the bot responses need to be saved on xr database so they can retrieve it when they want to . So I am now planning to work on that discussing with alex . The responses can be multiple according to & like if I say 'hi' to the agent it can reply like
'line 1 .......'
'line 2 ......'
'line 3 .....'
and also we have to map users and many things yet to come from & to work on this API .
but the basic requirement starts from this

review issue and pull request, upgrade all dependencies, make sure nothing is broken after
#101
#187

Discord does not allow normal bots to control user accounts. This also allows Monika to assist users in their DMs, whether by finishing a conversation for them, or by reading the message history and