interpreter 命令

常用示例

说明

Open Interpreter provides a natural language interface for computers, allowing language models to execute code (Python, JavaScript, Shell, and more) locally on your machine. It functions like ChatGPT's Code Interpreter but runs in your terminal with full access to your local environment. The tool equips function-calling language models with an exec() function that accepts a programming language and code to run. Since code executes in your local environment, it can interact with files, install packages, browse the web, and control system settings. By default, interpreter asks for confirmation before executing code. Open Interpreter supports multiple model providers including OpenAI, Anthropic, and local models via Ollama, LM Studio, or jan.ai. Configuration can be saved to YAML profiles for easy switching between setups.

参数

-y, --auto-run

Run code without manual confirmation at each step

--model _MODEL_

Specify which model to use (e.g., gpt-4o, claude-3-sonnet)

--local

Use a local model for code interpretation (via Ollama, LM Studio, jan.ai, etc.)

--fast

Use a faster, smaller model for quicker responses

--verbose

Activate verbose mode for detailed output

--os

Enable OS mode for controlling your computer (mouse, keyboard, screen)

--max_tokens _N_

Maximum tokens for model output

--context_window _N_

Context window size (must be larger than max_tokens)

--profile _FILE_

Load configuration from a YAML profile

--save-profile _NAME_

Save current configuration as a named profile

--tools _LIST_

Enable specific tools (default: interpreter,editor)

--stdin

Accept input from stdin for Unix-style piping

--custom_instructions _TEXT_

Append custom instructions to the system message (e.g., system info, preferred languages).

--api_base _URL_

Custom API base URL for model provider.

--api_key _KEY_

API key for authentication.