Gemini slackbot example

This is Gemini Pro integrated slack (bolt JavaScript framework) sample app. Users can have a conversation with Gemini.

deploy you're environment

[deprecated] One Click Deploy Button

Cloud Run one click deploy button is deprecated... Currently we are considering the next best deployment process. Now just intractive shell script.

deploy from your local

$ npm run deploy

Then intractive shell scripts are launched. Input followings

• Google Cloud Platform region
  • build and deploy target region
• Google Cloud Platform project name
  • target project ID not number
• Google Cloud service account for cloud run
• slack bot authentication
  1. slack signing secret
  2. slack oauth2 token for bot

Google Cloud Service Account

only one Google Cloud Permission

• "aiplatform.endpoints.predict"

Slack integration

1. Create slack app

• Go to https://api.slack.com/apps
• Click "Create New App" button
• Select "From an app manifest"
• Select target workspace
• Paste the slack app manifest from slack.yaml

2. Retrive slack authentications

a. slack signing secret

Go to the app page like "https://https://api.slack.com/apps/SLACK_APP_ID"

b. bot oauth2 token

Go to the auth page like "https://api.slack.com/apps/SLACK_APP_ID/oauth" and install the app your workspace.

After installing OAuth2 bot token will be appeared in the same page.

3. Set the event subscription URL

Replace "YOUR_CLOUD_RUN_ENDPOINT" after npm run deploy

local development

$ SLACK_SIGNING_SECRET=YOUR_OWN_ONE \
  SLACK_OAUTH2_BOT_TOKEN=YOUR_OWN_ONE \
  SLACK_SOCKET_MODE=true \
  SLACK_APP_TOKEN=YOUR_OWN_ONE \
  GCP_PROJECT=YOUR_OWN_ONE \
  GCP_REGION=YOUR_OWN_ONE \
  npm run start:dev

environment variables

optional

• LLM_MODEL : "gemini" or "claude3"