Configure OpenClaw’s Slack Transport

Use Socket Mode to connect OpenClaw to Slack when you want the default transport and have no public request endpoint. Choose HTTP Request URLs when you operate that endpoint, or relay mode when a trusted router owns Slack ingress.

Choose who receives Slack events

Choose Socket Mode when the OpenClaw operator owns the Slack app and can run the Gateway without publishing a request endpoint. OpenClaw uses Socket Mode by default. If your team already owns the public endpoint where Slack should send requests, use HTTP Request URLs and identify who will verify signing and troubleshoot incoming requests. For a managed deployment, use relay mode when a trusted router already receives Slack events.

Decide which conversation to test before you configure the app. Pick a DM for an operator-only pilot or one channel mention for a team pilot. Record the workspace, app, Gateway, and expected destination together. That gives everyone a concrete path to test when Slack events and OpenClaw replies have different owners.

Track ambiguous deliveries by event ID in your own logs. A socket connection and a posted reply are distinct observations.

Slack - OpenClaw

Test runtime status and delivery separately

In this illustrative setup, the Sandbox workspace uses Socket Mode and the pilot consists of one operator DM and one mention in agent-pilot. Run openclaw channels status --probe and inspect the Slack entry. If the entry reports a blocked lifecycle and a plugin error, repair that plugin issue and probe again before changing Slack event subscriptions. Record which workspace you installed the Slack app in before testing delivery.

In the Slack app's DM, ask the bot to reply with SLACK-DM-1. If pairing is requested, approve the expected sender through OpenClaw and send the message again. In agent-pilot, mention the app and ask for SLACK-CHANNEL-1. The pilot passes when both replies arrive in their intended conversations. A DM reply paired with channel silence narrows the investigation to the channel path.

Test with one synthetic event from the intended workspace. Record whether acknowledgment and response posting occurred on the same request or on separate jobs.

Channels - OpenClaw

Configure the app and the matching credentials

For Socket Mode, create a Slack app in the intended workspace and use the OpenClaw Slack manifest and scopes guide as your setup checklist. Enable Socket Mode, generate the app-level token, and install the app in the workspace to obtain its bot token. Put the two credentials in OpenClaw's corresponding appToken and botToken fields and apply the configuration to the Gateway that will receive events.

Enable the bot event subscriptions for the pilot conversation and enable the Slack app's messaging surface if the pilot uses DMs. For a channel pilot, include mention handling and add the app to the test channel. Confirm that the installed workspace matches the workspace you are testing before sending the first message. Add more event types when you have a specific workflow to test for them.

Slack - OpenClaw

Fix the event path that failed

When a Slack DM works but a channel mention fails, use the same operator and prompt in the intended pilot channel. Check the app's channel membership and mention subscription, then check the OpenClaw channel allowlist and mention gating. Keep the working credentials in place while making one change per test. A reply in the intended channel is the signal to stop changing settings.

If both tests fail on HTTP Request URLs, check that the Slack request URL reaches the configured webhookPath and that the signingSecret belongs to the same Slack app. Identify whether the receiver saw the event before inspecting downstream access policies. For relay mode, ask the router operator to confirm receipt and forwarding of the same test event, then check whether OpenClaw processed it.

Slack - OpenClaw

Sources