Chat Command

This guide documents the chat command in the Kubiya CLI, which allows you to interact with Kubiya teammates through your terminal.

Overview

The chat command enables conversational interaction with Kubiya AI teammates. It provides a natural language interface for asking questions, getting assistance, and executing tools through conversations.

Basic Usage

There are several ways to use the chat command:

# Interactive chat mode with a full terminal UI
kubiya chat --interactive

# Send a message to a specific teammate by ID
kubiya chat -i "teammate-id" -m "How can I deploy to Kubernetes?"

# Send a message to a specific teammate by name
kubiya chat -n "DevOps Expert" -m "Help me with Docker"

# Include files as context
kubiya chat -n "security" -m "Review this code" --context "src/*.go" 

# Pipe input from stdin
cat error.log | kubiya chat -n "debug" --stdin

Command Options

Option	Description
--interactive	Start an interactive chat session with a terminal UI
-i, --id	Specify the teammate ID to chat with
-n, --name	Specify the teammate name to chat with
-m, --message	Message to send (required unless using --interactive or --stdin)
--context	Files or URLs to include as context (supports wildcards)
--stdin	Read message from standard input
--stream	Stream the response as it's generated
--session-id	Continue a previous session using its ID
--clear-session	Clear the saved session
--debug	Enable debug logging

Context Files

You can include files or URLs as context for your chat:

# Using context files with wildcards
kubiya chat -n "security" -m "Review this code" --context "src/*.go" --context "tests/**/*_test.go"

# Using URLs as context
kubiya chat -n "security" -m "Check this" --context https://raw.githubusercontent.com/org/repo/main/config.yaml

# Multiple context sources
kubiya chat -n "devops" \
  --context "k8s/*.yaml" \
  --context "https://example.com/deployment.yaml" \
  --context "Dockerfile" \
  -m "Review deployment"

Interactive Mode

The interactive mode provides a full terminal UI for chatting:

kubiya chat --interactive

In interactive mode, you can:

• Chat continuously with an AI teammate

• View tool executions in real-time

• Clear the conversation

• Browse previous messages

• Export the conversation

Working with Sessions

Sessions allow you to continue conversations:

# Continue a previous session
kubiya chat -n "DevOps Expert" -m "Continue from before" --session-id "sess_12345"

# Clear the saved session
kubiya chat --clear-session

Streaming Responses

Use the --stream flag to see responses as they're generated:

kubiya chat -n "DevOps Expert" -m "Explain Kubernetes networking" --stream

Piping Input

You can pipe content from other commands or files:

# Pipe logs to chat
cat error.log | kubiya chat -n "debug" --stdin

# Pipe command output
kubectl get pods | kubiya chat -n "k8s" --stdin --context "k8s/deployment.yaml"

Example Use Cases

Troubleshooting

# Send logs for analysis
cat application.log | kubiya chat -n "Application Expert" --stdin

Code Review

# Get a code review
kubiya chat -n "Code Reviewer" -m "Review my implementation" --context "src/*.js"

Learning and Guidance

# Ask for guidance on a technology
kubiya chat -n "DevOps Coach" -m "How should I structure my Terraform files?"

Task Automation

# Describe what you want to accomplish
kubiya chat -n "Automation Expert" -m "Create a script to backup my database daily"

Troubleshooting

Common Issues and Solutions

1. Teammate Not Found

   • Verify the teammate name or ID

   • Use kubiya teammate list to see available teammates

2. Context Files Not Loading

   • Check file paths and permissions

   • Ensure wildcards are properly quoted

   • Verify URL accessibility

3. Session Issues

   • Use --clear-session to start fresh if encountering problems

   • Specify a new session-id if you want to start a new conversation

Related Commands

• kubiya teammate list - View available teammates

• kubiya tool - Manage automation tools

Conclusion

The chat command provides a flexible interface for interacting with Kubiya's AI teammates. Whether you need quick answers, detailed assistance with code, or help with complex tasks, the chat command offers a natural way to communicate and get results through your terminal.