Run Your Agent

Run your agent on your local computer by typing the following in your terminal:

ship run local

This command will boot a local server that provides a number of options to interact with your agent:

~/multimodal-agent-starter (main*) » ship run local                                                                                                             1 ↵ edwardbenson@Steamship
Steamship Python CLI version 2.17.8
Running your project...
 
📝 Log file:   logs/shiplog--2023-07-13--16:06:27.log
🌎 Public API: https://some-identifier.ngrok-free.app
🌎 Local API:  http://localhost:8443
🌎 Web UI:     https://steamship.com/debug?endpoint=https://some-identifier.ngrok-free.app/prompt
 
💬 Interactive REPL below. Type to interact.
 
Starting REPL for Agent...
If you make code changes, restart this REPL. Press CTRL+C to exit at any time.
 
Input: _

You can use the console to REPL to begin talking with your agent immediately, but you can also:

• Use the Web UI to interact via your browser
• Use the local API endpoint to interact programmatically
• Use the public API endpoint to test integrations (e.g. with Telegram or Slack)

Common Problems when Running

• I'm getting an error that the port is in use. Sometimes the development server gets stuck shutting down. Just re-run with another port number using the --port flag. For example: ship run local --port 8889. The original port will usually be released within about a minute.

• I'm getting an NGROK error. Sometimes VPNs or other network situations can cause complications with NGROK. Just run with the --no-ngrok flag to suppress the creation of a public URL. Note that this will prevent you from testing integrations with external webhooks.

• I'm getting an error about the Steamship manifest or api.py file. This may mean you're running the ship run local command from a directory other than the one containing your agent. Make sure you are running from your project directory and try again.