← 返回命令列表

Linux command

anthropic 命令

文本

复制后可按需替换文件名、目录或参数。

常用示例

Send a simple message

anthropic -g user "Hello, Claude!"

Send a message

anthropic -g user "[prompt]" -m [claude-3-5-sonnet-20241022]

Analyze an image

anthropic -g user "Describe this image" -i [path/to/image.png]

Set a system prompt

anthropic -g user "[prompt]" -s "You are a helpful coding assistant"

Multi-turn conversation

anthropic -g user "Hello!" -g assistant "Hi there!" -g user "How are you?"

Control response length

anthropic -g user "[prompt]" -x [2048] -t [0.7]

说明

anthropic is a command-line interface for interacting with Anthropic's Claude AI models. It provides direct access to the Claude API from the terminal, allowing users to send text messages and images for analysis. The tool supports multi-turn conversations by chaining multiple -g flags, enabling context-aware interactions. It handles various image formats including PNG, JPEG, and PDF files for vision tasks. The CLI integrates well with shell pipelines, making it useful for scripting and automation workflows.

参数

-g, --message _role_ _content_
Add a message with the specified role (user/assistant) and content. Can be repeated for multi-turn conversations.
-m, --model _model_
Specify the Claude model to use. Default: claude-3-opus-20240229.
-s, --system _prompt_
System message providing context or instructions for the conversation.
-i, --image _path_
Path to an image file (PNG, JPEG, or PDF) for vision tasks.
-t, --temperature _value_
Controls response randomness. Higher values produce more creative outputs.
-k, --top_k _value_
Restricts sampling to the top k most likely tokens.
-p, --top_p _value_
Nucleus sampling; restricts to tokens with cumulative probability above threshold.
-x, --max_tokens _count_
Maximum number of tokens in the response. Default: 1024.

FAQ

What is the anthropic command used for?

anthropic is a command-line interface for interacting with Anthropic's Claude AI models. It provides direct access to the Claude API from the terminal, allowing users to send text messages and images for analysis. The tool supports multi-turn conversations by chaining multiple -g flags, enabling context-aware interactions. It handles various image formats including PNG, JPEG, and PDF files for vision tasks. The CLI integrates well with shell pipelines, making it useful for scripting and automation workflows.

How do I run a basic anthropic example?

Run `anthropic -g user "Hello, Claude!"` in a terminal, then adjust file names, paths, flags, or remote targets for your system.

What does -g, --message _role_ _content_ do in anthropic?

Add a message with the specified role (user/assistant) and content. Can be repeated for multi-turn conversations.