Skip to main content

​
Prerequisites

Before using the PotpieAI integration for Slack, you need:
  • A PotpieAI account with an API token
  • At least one project that has been parsed and is in ready status
  • The PotpieAI app for Slack installed in your workspace

​
Installation

  1. Install the PotpieAI app for Slack
  2. Grant permissions when prompted
  3. PotpieAI appears in the workspace apps list

​
Authentication

Authenticate with an API token before sending the first message.
  1. Type /authenticate in any channel or direct message
  2. A modal opens asking for an API token
  3. Enter a PotpieAI API token. Follow the API access documentation to generate one.
  4. Click Submit
A confirmation message appears after successful authentication.

​
Using PotpieAI in Slack

​
Starting a Conversation

  1. Type /potpie in any channel or direct message
  2. A modal opens with the following options:
    • Repository selection from parsed repositories
    • Agent selection
    • Query input
  3. Fill in all fields and click Submit
PotpieAI creates a new thread and responds based on the codebase.

​
Continuing the Conversation

  1. Reply to the thread by @mentioning PotpieAI (e.g., @PotpieAI tell me more about this function)
  2. The agent processes the follow-up question within the ongoing conversation context

​
Command Reference

CommandDescription
/authenticateSet up a PotpieAI API token
/potpieStart a new conversation with an AI agent
@PotpieAI [question]Continue a conversation in a thread

​
Privacy Policy

Review the Privacy Policy for details on data collection and protection.

​
Disclaimer

AI-Generated Content Notice: PotpieAI uses Large Language Models (LLMs) to analyze the codebase and generate responses. AI agents strive to provide accurate and helpful information, but:
  • AI-generated responses may contain inaccuracies, errors, or misinterpretations
  • Always verify critical information and code suggestions before implementation
  • The AI’s understanding reflects the parsed codebase and may not match the current state of the code
  • Complex or ambiguous queries may result in incomplete or incorrect responses
Treat AI responses as suggestions. Review and validate all generated content.