将此文件放入 ~/.openinterpreter/config.toml 并根据你的工作流进行编辑。此处的每个键都是可选的。
# ---------------------------------------------------------------
# Model and provider
# ---------------------------------------------------------------
model_provider = "openai"
model = "gpt-5.1-codex"
model_reasoning_effort = "medium"
model_reasoning_summary = "auto"
personality = "pragmatic"
web_search = "cached"
# ---------------------------------------------------------------
# Harness compatibility
# ---------------------------------------------------------------
# Leave unset to choose the recommended harness for the selected provider.
# harness = "kimi-code"
harness_guidance = true
# ---------------------------------------------------------------
# Sandbox and approvals
# ---------------------------------------------------------------
sandbox_mode = "workspace-write"
approval_policy = "on-request"
[sandbox_workspace_write]
network_access = false
writable_roots = []
# ---------------------------------------------------------------
# Logging and history
# ---------------------------------------------------------------
log_dir = "~/.openinterpreter/log"
[history]
persistence = "save-all"
max_bytes = 104857600
# ---------------------------------------------------------------
# Profiles
# ---------------------------------------------------------------
[profiles.fast]
model = "gpt-5.1-codex-mini"
model_reasoning_effort = "low"
[profiles.review]
model = "gpt-5.1-codex"
model_reasoning_effort = "high"
sandbox_mode = "read-only"
# ---------------------------------------------------------------
# Custom provider
# ---------------------------------------------------------------
[model_providers.example]
name = "Example"
base_url = "https://api.example.com/v1"
env_key = "EXAMPLE_API_KEY"
wire_api = "responses"
# ---------------------------------------------------------------
# MCP servers
# ---------------------------------------------------------------
[mcp_servers.docs]
command = "docs-server"
default_tools_approval_mode = "prompt"
[mcp_servers.docs.tools.search]
approval_mode = "approve"
# ---------------------------------------------------------------
# Features
# ---------------------------------------------------------------
[features]
hooks = true
multi_agent = true
shell_tool = true
shell_snapshot = true
unified_exec = true
memories = false
apps = false
plugins = false
undo = false使用配置文件运行：
interpreter --profile review为单次运行覆盖一个值：
interpreter -c approval_policy='"never"' "fix the failing tests"