Skip to content

interserver/teams-chat-bot

BranchesTags

Repository files navigation

page_type sample
description This bot can capture all channel messages in Teams using RSC permissions, without the need for @mentions.
products
office-teams
office
office-365
languages
nodejs
extensions
contentType createdDate
samples
06/10/2021 01:48:56 AM
urlFragment officedev-microsoft-teams-samples-bot-receive-channel-messages-withRSC-nodejs

Receive Channel messages with RSC permissions

This sample app illustrates how a bot can capture all channel messages in Microsoft Teams by utilizing RSC (resource-specific consent) permissions, eliminating the need for @mentions. The bot supports adaptive card responses, easy local testing with tools like ngrok or dev tunnels, and deployment to Azure, allowing it to function effectively across different channels and group chats in Teams.

This feature shown in this sample is currently available in Public Developer Preview only.

Included Features

  • Bots
  • Adaptive Cards
  • RSC Permissions

Interaction with app

Bot Receive Channel MessagesWithRSCGif

Try it yourself - experience the App in your Microsoft Teams client

Please find below demo manifest which is deployed on Microsoft Azure and you can try it yourself by uploading the app package (.zip file link below) to your teams and/or as a personal app. (Uploading must be enabled for your tenant, see steps here).

Receive Channel messages with RSC permissions: Manifest

Prerequisites

  1. Office 365 tenant. You can get a free tenant for development use by signing up for the Office 365 Developer Program.

  2. To test locally, NodeJS must be installed on your development machine (version 16.14.2 or higher).

    # determine node version
node --version

  3. dev tunnel or Ngrok (For local environment testing) latest version (any other tunneling software can also be used)

    If you are using Ngrok to test locally, you'll need Ngrok installed on your development machine. Make sure you've downloaded and installed Ngrok on your local machine. ngrok will tunnel requests from the Internet to your local computer and terminate the SSL connection from Teams.

  4. Microsoft 365 Agents Toolkit for VS Code or TeamsFx CLI

Run the app (Using Microsoft 365 Agents Toolkit for Visual Studio Code)

The simplest way to run this sample in Teams is to use Microsoft 365 Agents Toolkit for Visual Studio Code.

  1. Ensure you have downloaded and installed Visual Studio Code
  2. Install the Microsoft 365 Agents Toolkit extension
  3. Select File > Open Folder in VS Code and choose this samples directory from the repo
  4. Using the extension, sign in with your Microsoft 365 account where you have permissions to upload custom apps
  5. Select Debug > Start Debugging or F5 to run the app in a Teams web client.
  6. In the browser that launches, select the Add button to install the app to Teams.

If you do not have permission to upload custom apps (uploading), Microsoft 365 Agents Toolkit will recommend creating and using a Microsoft 365 Developer Program account - a free program to get your own dev environment sandbox that includes Teams.

Setup

NOTE: The free ngrok plan will generate a new URL every time you run it, which requires you to update your Azure AD registration, the Teams app manifest, and the project configuration. A paid account with a permanent ngrok URL is recommended.

  1. Setup for Bot

  • Register Azure AD application

  • Register a bot with Azure Bot Service, following the instructions here.

  • Ensure that you've enabled the Teams Channel

  • While registering the bot, use https://<your_tunnel_domain>/api/messages as the messaging endpoint.

    NOTE: When you create your app registration in Azure portal, you will create an App ID and App password - make sure you keep these for later.

  1. Setup NGROK

  • Run ngrok - point to port 3978

    ngrok http 3978 --host-header="localhost:3978"

    Alternatively, you can also use the dev tunnels. Please follow Create and host a dev tunnel and host the tunnel with anonymous user access command as shown below:

    devtunnel host -p 3978 --allow-anonymous
  1. Setup for code

  • Clone the repository

    git clone https://github.com/OfficeDev/Microsoft-Teams-Samples.git

  • In the folder where repository is cloned navigate to samples/bot-receive-channel-messages-withRSC/nodejs

  • Install node modules

    Inside node js folder, open your local terminal and run the below command to install node modules. You can do the same in Visual Studio code terminal by opening the project in Visual Studio code.

    npm install

  • Update the .env configuration for the bot to use the MicrosoftAppId (Microsoft App Id) and MicrosoftAppPassword (App Password) from the Microsoft Entra ID app registration in Azure portal or from Bot Framework registration.

NOTE: the App Password is referred to as the client secret in the azure portal app registration service and you can always create a new client secret anytime.

  • Run your app

    npm start

  • Install modules & Run the NodeJS Server

    • Server will run on PORT: 3978
    • Open a terminal and navigate to project root directory
      npm run server

NOTE:This command is equivalent to: npm install > npm start

  1. Run your app

    npm start

  2. Setup Manifest for Teams

    • Edit the manifest.json contained in the appManifest folder to replace your Microsoft App Id (that was created when you registered your bot earlier) everywhere you see the place holder string <<YOUR-MICROSOFT-APP-ID>> (depending on the scenario the Microsoft App Id may occur multiple times in the manifest.json) <<DOMAIN-NAME>> with base Url domain. E.g. if you are using ngrok it would be https://1234.ngrok-free.app then your domain-name will be 1234.ngrok-free.app and if you are using dev tunnels then your domain will be like: 12345.devtunnels.ms. Replace <> with any GUID or with your MicrosoftAppId/app id

    • Zip up the contents of the appManifest folder to create a manifest.zip

    • Upload in a team to test

      • Select or create a team
      • Select the ellipses ... from the left pane. The drop-down menu appears.
      • Select Manage Team, then select Apps
      • Then select Upload a custom app from the lower right corner.
      • Then select the manifest.zip file from appManifest, and then select Add to add the bot to your selected team.

Note: If you are facing any issue in your app, please uncomment this line and put your debugger for local debug.

Running the sample

Adding bot UI:

App installation

Hey command interaction:

Permissions

1 or 2 command interaction:

Permissions

Adding App to group chat:

Adding To Groupchat

Group chat interaction with bot without being @mentioned:

Group Chat

Interacting with the bot in Teams

Select a channel and enter a message in the channel for your bot.

The bot receives the message without being @mentioned.

Deploy the bot to Azure

To learn more about deploying a bot to Azure, see Deploy your bot to Azure for a complete list of deployment instructions.

Further reading

Changelog

v2 — Notification Queue, Daily Recap Cards, Architecture Modernisation

This release adds a Redis-backed notification queue consumer, a daily recap Adaptive Card system, and a major architectural cleanup (shared adapter, shared Redis factory, rate limiting, auth).

New Features

Daily Recap Card API (POST /api/dailyrecap)

  • Proactively post Adaptive Cards built from Adaptive Cards Templating template + data pairs
  • Server-side binding via adaptivecards-templating SDK — Teams cannot expand ${...} templating syntax client-side
  • X-Daily-Recap-Token header validated against DAILY_RECAP_TOKEN env var
  • Auto-deletes the sent card after 5 minutes to avoid channel clutter
  • Caches template+data in Redis (dailyrecap:{activityId}, 2-day TTL) for later Action.Submit toggle edits
  • Card size enforcement: cards over 25 KB are trimmed (text truncation, base64 image removal, empty-array pruning); if still too large the request is rejected with 400
  • Response: { ok, activityId, conversationId, channel, bytes_sent }

daily recap Bot Command (admin-only)

  • Fetches the card template+data from DAILY_RECAP_URL and posts it to the current channel
  • Compact mode trims the card so it stays under Teams' 25 KB Adaptive Card limit
  • Auto-deletes after 5 minutes

Notification Queue Consumer (server/queue/notificationConsumer.js)

  • Polls Redis (notif:queue) on a configurable interval; speeds up to NOTIF_POLL_FAST_MS after activity
  • Per-tick batch limit (NOTIF_MAX_PER_TICK, default 50)
  • Three send modes per room:
    • Trackable (envelopes with extra.dedup_key): edits the most recent activity within a 30-minute window; appends a timestamped line up to 3 times, then summarises as "N updates · last HH:MM"
    • Coalesced (2–8 non-deduped messages): combined into one activity, separated by — · — dividers
    • Single (one non-deduped message): sent as a new activity
  • Edit coalescing: messages with the same dedup_key within NOTIF_EDIT_WINDOW_MS (default 30 min) update the same activity instead of spawning new ones
  • Fallback to Power Automate webhook: if adapter.continueConversation fails after retries, the envelope's fallback_webhook_url is used
  • Dead-letter queue (notif:dead): failed/expired envelopes stored with _dead_reason and _dead_at; !notif drain-dead re-queues them
  • GitHub noise filtering (server/queue/filters.js): check_run, workflow_job, check_suite, workflow_run, star, watch, fork, ping events are silently dropped or redirected to int-dev-announce with a visible label; control via NOTIF_FILTER_ENABLED=0
  • Recovery on startup: any items left in notif:processing (from a crashed tick) are moved back to notif:queue
  • Metrics counters: notif:metrics:enqueued, sent, edited, coalesced, redirected, fallback, dead
  • Heartbeat logging every NOTIF_HEARTBEAT_MS (default 60s) when the queue is idle

!notif Admin Commands (via server/commands/notifAdmin.js)

  • !notif status — queue/processing/dead depths + all metric counters
  • !notif rooms — table of known rooms; ✅ = convref cached, ❌ = bot needs to observe inbound activity first
  • !notif test <room> <msg> — enqueues a probe envelope (dedup_key: admin:test); message capped at 4000 chars
  • !notif drain-dead — moves all entries from notif:dead back to notif:queue
  • !notif seed-room <room> — reports whether a conversation reference exists for the named room; room name capped at 64 chars

Channel Sync on Startup

  • Set CHANNEL_SYNC_ENABLED=1 to probe every channel in CHANNELS on startup
  • Channels without a stored convref receive an invisible zero-width-space message (\u200b) which is deleted after 500ms — registers the bot's presence without polluting the channel
  • The onMessage handler already stores convref:* keys on every inbound message; sync fills gaps for channels added after deployment

Constructed ConversationReference Fallback

  • When no convref is stored for a room (e.g. the bot was never observed there), tryConstructedConvRef attempts the send using a minimal reference with the known conversationId
  • Uses configurable TEAMS_SERVICE_URL env var (supports GCC/GCC High if the correct URL is set)
  • Falls back to the envelope's fallback_webhook_url if even the constructed reference fails

Architecture / Refactoring

Shared BotFrameworkAdapter (server/lib/adapter.js)

  • Single adapter singleton shared by botController, msgController, dailyRecapController, notificationConsumer, and botActivityHandler
  • Credentials and token cache are pooled; onTurnError handled per-module where needed
  • setAdapter() allows test injection

Shared Redis Factory (server/lib/redis.js)

  • createBotRedis() — for convref:* keys; supports REDIS_USER/REDIS_PASSWORD auth; maxRetriesPerRequest: 3
  • createNotifRedis() — for the notification queue; separate host via REDIS_HOST_MY/REDIS_PORT_MY
  • TEAMS_SERVICE_URL — single source of truth for the Teams endpoint (default: https://smba.trafficmanager.net/teams/); overridden by env var for GCC/GCC High

Shared Retry Library (server/lib/retry.js)

  • Exponential backoff with jitter (base 750ms, cap 8s) replacing inline retry loops
  • Classifies errors as transient, auth, or rateLimit; re-trusts serviceUrl on auth failures
  • Used by all outbound Bot Framework calls across the codebase

Channel-Based Routing

  • msgController.js now accepts channel in the request body and resolves it via CHANNELS map
  • Hardcoded int-dev-private target removed; the caller specifies the channel
  • SKIP_CHANNELS array disables specific channels without code changes

Rate Limiting

  • express-rate-limit added as a dependency
  • General /api limit: 60 req/min per IP (RATE_LIMIT_WINDOW_MS, RATE_LIMIT_MAX)
  • Proactive-send endpoints (/api/message, /api/dailyrecap): 30 req/min per IP+channel (RATE_LIMIT_SEND_WINDOW_MS, RATE_LIMIT_SEND_MAX)

Infrastructure

New Environment Variables

Variable Default Purpose
DAILY_RECAP_URL https://mystage.interserver.net/admin/ajax/daily_recap_card_api.php MyAdmin daily recap endpoint
DAILY_RECAP_TOKEN (none) Shared secret for daily recap auth
REDIS_HOST_MY / REDIS_PORT_MY 67.217.60.234 / 6379 Notification queue Redis
REDIS_USER / REDIS_PASSWORD (none) Bot Redis authentication
TEAMS_SERVICE_URL https://smba.trafficmanager.net/teams/ Teams API endpoint (change for GCC/GCC High)
NOTIF_POLL_MS 5000 Queue poll interval (ms)
NOTIF_POLL_FAST_MS 1000 Poll interval after activity (ms)
NOTIF_MAX_PER_TICK 50 Max envelopes per poll
NOTIF_COALESCE_MAX_CHARS 24000 Max chars in a coalesced message
NOTIF_COALESCE_MAX_ITEMS 8 Max items in a coalesced message
NOTIF_EDIT_WINDOW_MS 1800000 (30 min) Dedup edit window
NOTIF_KEY_PREFIX notif: Redis key prefix
NOTIF_HEARTBEAT_MS 60000 Heartbeat log interval (ms)
NOTIF_FILTER_ENABLED 1 Set 0 to bypass GitHub noise filters
NOTIF_CONSUMER_ENABLED 1 Set 0 to disable notification consumer
CHANNEL_SYNC_ENABLED 0 Set 1 to probe channels on startup
RATE_LIMIT_WINDOW_MS 60000 General rate limit window
RATE_LIMIT_MAX 60 General rate limit max reqs
RATE_LIMIT_SEND_WINDOW_MS 60000 Send-endpoint rate limit window
RATE_LIMIT_SEND_MAX 30 Send-endpoint rate limit max reqs
NOTIF_COMMIT_GROUP_WINDOW_MS 180000 (3 min) GitHub commit grouping window
NOTIF_DOWNSTREAM_REPOS detain/sugarcraft:sugarcraft/* Comma-separated upstream:glob pairs. Pushes to a repo matching glob are treated as action-triggered and attributed to the most recent active-CI parent in upstream

GitHub Commit + Job Grouping

  • check_run and workflow_job events are no longer silently dropped — they are passed through to the notification consumer.
  • A dedup_key of github:commit:{sha} (first 7 chars) is auto-injected for GitHub events that carry a commit SHA and have no pre-existing dedup_key.
  • All events for the same commit SHA are routed to the same trackable message within the edit window (30 min).
  • The resulting message has the commit header (SHA + first line of commit message) followed by a — jobs — section listing each job's name, failures, status/conclusion, elapsed time, and URL.
  • Subsequent job statuses for the same commit edit that same message, updating the job's line in-place as it progresses: queued → in_progress → success/failure.
  • If no event for a given SHA arrives within 3 minutes (NOTIF_COMMIT_GROUP_WINDOW_MS), the edit-window cache expires and the next event for that SHA starts a fresh message.
  • check_suite and workflow_run remain silently filtered (they are aggregates; individual job events carry the detail).

Action-triggered Push Attribution

  • Workflow events seed a per-repo active-CI index (notif:wfactive:{repo} sorted set, member = parent SHA, score = ts) on every workflow_run / workflow_job / check_run arrival.
  • When a push event looks action-triggered — pusher is github-actions[bot] / matching github-actions@, sender ends in [bot], head commit author is a github-actions identity, OR the repo lives under a NOTIF_DOWNSTREAM_REPOS glob — the consumer looks up the most recent active parent SHA (in the same repo first, then in any upstream repo via the map) and rewrites dedup_key to that parent's. The push then nests under the existing trackable instead of spawning a new message.
  • Default downstream map is detain/sugarcraft:sugarcraft/*, covering both the vhs.yml self-commit (bot pusher in same repo) and sync-sugarcraft.yml subtree pushes (PAT pusher to downstream repos).
  • !notif wfactive shows the current per-repo parents-with-active-CI list and the configured downstream map.

/health/queue Endpoint

  • Returns { status, running, queue_depth, processing_depth, dead_depth, poll_interval_ms, edit_window_ms, max_per_tick, last_tick }
  • Useful for monitoring dashboards and alerting

.logs/ in .gitignore

  • Local log output directories are now ignored

About

Chat bot for MS Teams to receive notifications from and interact with our Datacenter Administration and BIlling Automation

Resources

Readme

License

MIT license
Activity
Custom properties

Stars

1 star

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages