Swordfish

Including Events in Chat

The Events panel shows console logs, network requests, and browser interactions. Include them in your conversation to give Swordfish context.

What Are Events?

  • Console Logs - Errors, warnings, and messages from JavaScript
  • Network Requests - API calls and resource loads
  • Browser Interactions - Clicks and typing (when recording is enabled)

Including Events Manually

Each event in the Events panel has an "Include in conversation" button. Click it to:

  • Add the event to the conversation history
  • Make it visible to the agent in the chat
  • Give the agent context about what happened
Try it: Console Log
id: console_log_42
ERROR
TypeError: Cannot read property 'map' of undefined
Click "Include in conversation" to see the button change, and click the expand button (▶) to see details.
Try it: Network Request
id: network_request_15
POST 200 - SUCCESS
https://api.example.com/users
Click "Include in conversation", expand the request (▶), and then expand individual sections to see details.
💡 Tip

Including events is especially useful when debugging. For example, if you see an error in the console, include it in the conversation and ask the agent to fix it.

Agent Can Search Events Too

You don't always need to manually include events. The agent has tools to search through historical events:

  • get_console_logs - Search console logs by pattern, level, or time
  • get_network_requests - Search network requests by URL, method, status, or response content

For example, you can ask:

  • "Check the console logs for any errors"
  • "What network requests failed?"
  • "Search console logs for 'authentication'"

    • Note

    The agent can search through all events that have occurred during the conversation, not just the ones you manually include. Manual inclusion is useful when you want to highlight something specific or provide context upfront.

    When to Include vs. Let Agent Search

    Scenario Approach
    You see a specific error you want fixed Include it manually for immediate context
    You want the agent to investigate all errors Let the agent search: "Check console logs for errors"
    You notice a failed API call Include it manually to show what went wrong
    You want to understand all network activity Let the agent search: "What network requests were made?"

    Event Details

    When you include an event, the agent receives:

    • Console Logs: Level, message, source, URL, line number, stack trace (if available)
    • Network Requests: Method, URL, status, headers, request/response body, timing

    This detailed information helps the agent understand what happened and how to fix issues.