Struct Config

pub struct Config {

    pub system_prompt: Option<String>,
    pub model: Option<String>,
    pub mcp_config_path: Option<PathBuf>,
    pub allowed_tools: Option<Vec<String>>,
    pub stream_format: StreamFormat,
    pub non_interactive: bool,
    pub verbose: bool,
    pub max_tokens: Option<usize>,
    pub timeout_secs: Option<u64>,
    pub continue_session: bool,
    pub resume_session_id: Option<String>,
    pub append_system_prompt: Option<String>,
    pub disallowed_tools: Option<Vec<String>>,
    pub max_turns: Option<u32>,
    pub skip_permissions: bool,
    pub security_level: SecurityLevel,
}

Configuration options for Claude AI client

The Config struct holds all configuration options for the Claude AI client, including model selection, system prompts, tool permissions, and output formatting.

Examples

use claude_sdk_rs_core::{Config, StreamFormat};
use std::path::PathBuf;

// Default configuration
let config = Config::default();

// Custom configuration with builder pattern
let config = Config::builder()
    .model("claude-3-opus-20240229")
    .system_prompt("You are a helpful Rust programming assistant")
    .stream_format(StreamFormat::Json)
    .timeout_secs(60)
    .allowed_tools(vec!["bash".to_string(), "filesystem".to_string()])
    .build();

Fields

system_prompt: Option<String>

Optional system prompt to set the assistant’s behavior and context

This prompt is sent with every request to provide consistent context and instructions to Claude about how it should respond.

model: Option<String>

Claude model to use for requests

Available models include:

• claude-3-opus-20240229 - Most capable model
• claude-3-sonnet-20240229 - Balanced performance and cost
• claude-3-haiku-20240307 - Fastest and most cost-effective

If not specified, Claude CLI will use its default model.

mcp_config_path: Option<PathBuf>

Path to Model Context Protocol (MCP) configuration file

MCP allows Claude to interact with external tools and data sources. This should point to a valid MCP config file containing server definitions.

allowed_tools: Option<Vec<String>>

List of tools that Claude is allowed to use

Tools are specified using the format server_name__tool_name for MCP tools or simple names like bash for built-in tools. An empty list or None means all available tools are allowed.

Examples

let tools = vec![
    "bash".to_string(),
    "filesystem".to_string(),
    "mcp_server__database_query".to_string(),
];

stream_format: StreamFormat

Output format for Claude CLI responses

• Text: Plain text output (default)
• Json: Structured JSON with metadata
• StreamJson: Line-delimited JSON messages for streaming

non_interactive: bool

Whether to run Claude CLI in non-interactive mode

When true (default), Claude CLI won’t prompt for user input, making it suitable for programmatic use.

verbose: bool

Enable verbose output from Claude CLI

When true, additional debugging information will be included in the CLI output. Useful for troubleshooting.

max_tokens: Option<usize>

Maximum number of tokens to generate in the response

If not specified, Claude will use its default token limit. Setting this can help control response length and costs.

timeout_secs: Option<u64>

Timeout in seconds for Claude CLI execution (default: 30s)

How long to wait for Claude CLI to respond before timing out. Increase this for complex queries that might take longer to process.

continue_session: bool

Whether to continue the last session

When true, adds the --continue flag to resume the most recent conversation session. This allows for conversation continuity across multiple API calls.

resume_session_id: Option<String>

Session ID to resume

When set, adds the --resume flag with the specified session ID to continue a specific conversation session. This allows for resuming conversations from a particular point.

append_system_prompt: Option<String>

Additional system prompt to append

When set, adds the --append-system-prompt flag to extend the existing system prompt. Cannot be used together with system_prompt.

disallowed_tools: Option<Vec<String>>

List of tools that Claude is NOT allowed to use

Tools are specified using the same format as allowed_tools. This provides fine-grained control over tool restrictions.

max_turns: Option<u32>

Maximum number of conversation turns

When set, limits the conversation to the specified number of back-and-forth exchanges. Useful for controlling conversation length.

skip_permissions: bool

Whether to skip permission prompts (default: true)

When true, adds the --dangerously-skip-permissions flag to bypass tool permission prompts. This is enabled by default for programmatic use but can be disabled for additional security.

security_level: SecurityLevel

Security validation strictness level

Controls how strictly user input is validated for potential security threats.

• Strict: Blocks most special characters and patterns
• Balanced: Context-aware validation (default)
• Relaxed: Only blocks obvious attack patterns
• Disabled: No security validation (use with caution)

Implementations

impl Config

pub fn builder() -> ConfigBuilder

Create a new configuration builder

The builder pattern provides a fluent interface for creating configurations with custom settings.

Examples

use claude_sdk_rs_core::{Config, StreamFormat};

let config = Config::builder()
    .model("claude-3-opus-20240229")
    .system_prompt("You are a helpful assistant")
    .stream_format(StreamFormat::Json)
    .timeout_secs(120)
    .build();

pub fn validate(&self) -> Result<(), Error>

Validate the configuration

Checks all configuration values for validity according to defined limits and constraints. Returns an error if any validation fails.

Errors

Returns Error::InvalidInput if:

• System prompt exceeds maximum length
• Timeout is outside valid range
• Max tokens exceeds limit
• Tool names are invalid

Trait Implementations

impl Clone for Config

fn clone(&self) -> Config

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Config

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for Config

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for Config

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for Config

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.