Slack Integration
Slack is one way to talk to your personal Puck. Mention @Amp in a channel or thread to ask a question, start work, or manage an existing Amp thread without leaving Slack.
Install Amp for Slack
Use these steps if you are not signed in to Amp:
- Select Add to Slack above.
- Sign in to your Amp account. If you do not have an account, select Sign Up and create one. Amp keeps Workspace Integrations as your destination while you sign in or create your account.
- Create or join an Amp workspace if prompted. After you finish workspace setup, Amp opens Workspace Integrations.
- Make sure you are an admin of the Amp workspace and that Slack allows you to install apps in the Slack workspace. If you do not have either permission, ask the right admin to complete the installation.
- Find Slack in Workspace Integrations and select Connect a Slack Workspace.
- Review the requested permissions in Slack, choose the Slack workspace, and authorize Amp.
- Ask each Amp workspace member to open Personal Integrations and connect Slack. Each member must link their own Slack identity before they can talk to Puck in Slack.
After Slack authorizes Amp, Amp shows a Slack Connected confirmation with the next steps. Add @Amp to a Slack channel, then mention @Amp in a channel or thread to start work.
What You Can Do
Puck can use messages, threads, and attached images as context. Here are some ways people use it:
Conversations and Replies
The first mention starts a Puck conversation for that Slack thread. Puck replies in the same thread so the discussion and results stay together.
Mention @Amp again when you want to give Puck another instruction. Open the linked Amp thread when you want to inspect the complete conversation or continue in the app.
Notices
A paid Slack plan is required to use Amp in Slack’s AI agent app container. On Slack Free, users
can still mention @Amp in channels and threads, receive threaded replies, and get supported Amp
link previews.
Amp uses large language models and can generate inaccurate responses, summaries, code, findings, or other outputs. Review its work and test code before relying on it or taking production action.
Amp does not train on your data. Read the Privacy Policy for more information.