Skip to main content
The @checksum Slack bot lets your team trigger test generation from Slack — the same capability as /checksum generate on a GitHub pull request, without leaving the channel.
This is separate from Notification Connectors (webhooks that push health reports and run results into Slack). You can use both: connectors for outbound alerts, the bot for inbound @checksum commands.

What it does

When someone mentions @checksum in a channel where the app is installed:
  1. Checksum reads the thread context (the mention plus recent messages in that thread)
  2. A generation agent session starts against your Checksum project
  3. The bot replies in the thread with a link to track the session in the web app
  4. When generation completes, the agent delivers tests the same way as other generation flows (typically a PR to your tests repository when the session has pull-request context)
Example mentions:
The bot uses conversation context — not just the mention line — so replies in the thread before @checksum are included in the agent instructions.

Setup

Workspace admin access is required. The Slack section is under Settings → Integrations.
  1. Open Settings → Integrations in the Checksum web app and find the Slack card
  2. Click Add to Slack
  3. Complete the Slack OAuth flow and choose the workspace
  4. Invite the Checksum app to the channels where the team should use @checksum
After connection, the settings page shows the connected workspace name and team ID.
Only install the bot in Slack workspaces your organization owns. The bot token is scoped to channels the app is invited to.

Usage tips

The bot does not replace Notification Connectors — configure those separately under Settings → Integrations (Notification Connectors section) for automated health reports, bug alerts, and run summaries.

Slack notifications (webhooks vs bot)

Checksum supports two Slack integration styles: When both are configured, notification messages use distinct formatting per event type (health reports, bug detected, auto-heal batches, and so on) so channels can tell them apart at a glance.

Next Steps