Skip to main content
The Slack integration enables your AI agents to communicate with your team directly in Slack. Agents can respond to messages, reply in threads, send rich formatted messages, and be triggered by @mentions in channels.

​
Features

Direct Messages

Chat with Cotool directly via DM. The bot creates a personal chat for each conversation.

Channel Triggers

@mention the bot in channels to trigger agents with specific capabilities.

Emoji Reactions

React to any message with a configured emoji to run an agent on that message.

Thread Conversations

Agents automatically reply in threads to keep channels organized.

Rich Messages

Send formatted messages with buttons, charts, and file attachments.

​
Setup

1

Navigate to Integrations

Go to Settings > Integrations in Cotool and find the Slack integration.
2

Install the Slack App

Click Connect to begin the OAuth flow. You’ll be redirected to Slack to authorize the Cotool app.
3

Grant Permissions

Review and approve the requested permissions. The app needs access to:
  • Read messages in channels where it’s mentioned
  • Read and respond to direct messages
  • Post messages and reactions
  • Read the workspace’s custom emoji (so emoji reaction triggers can be picked from a list)
  • Upload files
4

Select Workspace

Choose which Slack workspace to connect. You can connect multiple workspaces if needed.

​
How It Works

​
Direct Messages

You can start a conversation with Cotool by sending a direct message to the bot:
  1. Find Cotool in your Slack workspace’s Apps section
  2. Send a message describing what you need help with
  3. The bot reacts with 👀 to acknowledge receipt
  4. A personal chat is created in Cotool and the agent processes your request
  5. The agent replies directly in the Slack thread with its findings
DM conversations create personal chats in Cotool. You can continue the conversation by replying in the same thread, and the agent will have full context of previous messages.

​
Channel @Mentions (Agent Triggers)

For channel interactions, you need to set up a Slack trigger on an agent:
1

Create or Edit an Agent

Navigate to the agent you want to trigger from Slack.
2

Add a Slack Trigger

In the agent’s Triggers tab, click Add Trigger and select Slack.
3

Configure Access (Optional)

Optionally restrict which users, user groups, or channels can trigger the agent.
4

@Mention in Slack

In any channel where the bot is present, @mention the bot with your request:
When you @mention the bot:
  1. The bot reacts with 👀 to acknowledge it received your message
  2. The configured agent processes your request
  3. The agent replies in a thread under your message with its findings
Unlike DMs, channel @mentions require a configured agent trigger. If no trigger is configured, the bot will let you know how to set one up.

​
Emoji Reactions (Agent Triggers)

An agent can also be triggered by adding an emoji reaction to a message — useful when you spot something worth investigating and don’t want to retype it.
1

Configure the Emoji

On the agent’s Slack trigger, under Which emoji reactions trigger this agent?, pick the emoji that should trigger it. The picker searches your workspace’s custom emoji alongside the standard set, so you can search by name or keyword (typing alarm finds 🚨). There is no separate on/off switch: picking at least one emoji enables reaction triggering, and leaving the field blank disables it.
2

React in Slack

Add one of those reactions to any message in a channel the trigger allows.
When someone adds a configured reaction:
  1. The bot reacts with 👀 to acknowledge it
  2. The agent runs with that specific message as its subject, and can read the surrounding thread for context
  3. If the message is already in a thread, the agent replies in that thread. If it isn’t, the agent’s reply starts a new thread on the message
The trigger’s channel and user allow lists apply to the person who added the reaction, exactly as they do for @mentions. If the channel is allowed but the reactor isn’t, the bot explains why nothing happened (once per message and emoji) rather than running the agent. In a channel the trigger doesn’t cover, the reaction is ignored silently — the bot won’t post in channels it wasn’t configured for.
A few behaviors worth knowing:
  • One run per message and emoji. If several people add the same reaction to the same message, or someone removes and re-adds it, the agent runs once — not once per person.
  • Reactions continue an existing conversation. If the thread already has a Cotool chat, reacting adds to that chat instead of starting a new one.
  • Cotool’s own reactions never trigger anything, including its 👀 acknowledgement.
  • Reactions are only honored for emoji configured on a Slack trigger. Because each channel can have only one Slack trigger, all the emoji configured for a channel route to that trigger’s agent.
  • There is a cap of 30 reaction-triggered runs per 30 minutes per trigger — the same budget an agent already has for its own runs, so a single trigger is no more limited than before. Beyond it, reactions are skipped (no 👀) and the bot posts one message in the thread explaining why. Hitting this almost always means an emoji is configured more broadly than intended — narrow the emoji list, or the channels, on the trigger. Runs resume automatically. The agent’s own overall run rate limit applies too, and is reported the same way.
Emoji reactions require the reaction_added bot event to be subscribed in the Slack app. This is already configured for the Cotool-hosted Slack app. If your organization uses its own custom Slack app, an admin must add reaction_added to its Event Subscriptions.

​
Thread Replies

All agent responses are sent as thread replies to keep channels organized. If you reply to an existing thread where Cotool has responded, the agent will:
  1. Re-enter the existing conversation with full context
  2. Process your follow-up message
  3. Reply in the same thread
This enables natural back-and-forth conversations without cluttering the main channel.

​
Available Tools

When an agent has the Slack integration enabled, it can use these tools:

​
Example Use Cases

Create an agent that monitors your SIEM and sends formatted alerts to a #security-alerts channel with relevant context and recommended actions.
Set up a trigger so on-call engineers can @mention the bot to quickly investigate incidents, gather context from multiple tools, and summarize findings.
Build an agent that processes access requests posted in a channel, checks user permissions, and either approves or escalates based on policy.
Schedule an agent to post daily summaries of open tickets, pending alerts, or team metrics to keep everyone informed.

​
Best Practices

Use descriptive agent names - When multiple agents have Slack triggers, clear names help users know which agent to @mention.
Restrict channel access - Use trigger configuration to limit which channels can trigger sensitive agents, preventing accidental exposure in public channels.
Leverage thread context - Agents automatically read the full thread history, so users can have multi-turn conversations without repeating context.