Ask Command

The ask command provides single-turn question answering without maintaining a session.

Usage

localgpt ask [OPTIONS] <QUESTION>

Options

Option	Description
-m, --model <MODEL>	Override the default model
-f, --format <FORMAT>	Output format: text (default) or json
--no-memory	Disable memory context loading
--no-tools	Disable tool execution

Examples

Basic Question

localgpt ask "What is the capital of France?"
# Paris is the capital of France.

JSON Output

localgpt ask --format json "List 3 programming languages"

Output:

{
  "response": "1. Python\n2. JavaScript\n3. Rust",
  "model": "gpt-4",
  "tokens": {
    "prompt": 45,
    "completion": 12
  }
}

Using a Specific Model

localgpt ask -m claude-3-sonnet "Explain quantum computing briefly"

Piping Input

echo "Summarize this" | localgpt ask -
cat error.log | localgpt ask "What's wrong with this log?"

With Memory Context

By default, ask loads relevant memory context:

# This will include relevant memory about your projects
localgpt ask "What was I working on last week?"

Disable memory loading for faster responses:

localgpt ask --no-memory "What is 2+2?"

Use Cases

The ask command is ideal for:

• Quick questions - Get an answer without starting a session
• Scripting - Use in shell scripts with JSON output
• Piping - Process output from other commands
• One-off tasks - Translation, summarization, etc.

Comparison with Chat

Feature	ask	chat
Multi-turn	No	Yes
Session persistence	No	Yes
Memory context	Optional	Always
Tool execution	Optional	Always
Interactive	No	Yes

Exit Codes

Code	Meaning
0	Success
1	Error (API, config, etc.)