Struct AgentConfig 

pub struct AgentConfig {
    pub max_iterations: usize,
    pub max_retries: u32,
    pub retry_delay_ms: u64,
    pub max_tool_result_size: usize,
    pub max_todo_reminders: usize,
    pub iteration_warning_threshold: usize,
    pub compression: CompressionConfig,
    pub max_tokens: u32,
    pub context_size_override: Option<u32>,
    pub think: bool,
    pub verify_strategy: VerificationStrategy,
    pub project_path: Option<PathBuf>,
}

Agent runtime configuration.

Controls iteration limits, retry behavior, size thresholds, and compression settings. All defaults match the original hardcoded values for backward compatibility.

Fields

max_iterations: usize

Maximum number of iterations in the main loop.

Default: 200

Why this value?

• Sufficient for most common tasks (file edits, code review, simple builds)
• Prevents infinite loops and runaway operations
• Balances task completion with resource efficiency

What happens when limit is reached?

• Agent stops execution gracefully
• User receives detailed warning message
• Can continue, break down task, or resume

Examples:

• Simple task (edit file): ~5-10 iterations
• Medium task (refactor module): ~15-30 iterations
• Complex task (build system): ~40-50 iterations (may hit limit)

max_retries: u32

Maximum retry attempts for API calls.

Default: 5

Why this value?

• Handles transient network failures
• Balances resilience with response time
• Most failures resolve within 2-3 retries

retry_delay_ms: u64

Delay between retry attempts (milliseconds).

Default: 1000ms

Why this value?

• Gives API server time to recover
• Not too long to frustrate users
• Often used with exponential backoff

max_tool_result_size: usize

Maximum size for tool execution results (bytes).

Default: 50,000 (50KB)

Why this value?

• Prevents oversized responses from breaking API calls
• Fits within typical context window limits
• Most tool results are <10KB

Truncation behavior:

• Results larger than this are truncated
• Truncation message added: “⚠️ Output truncated (X bytes total)”
• User can request specific portions if needed

max_todo_reminders: usize

Maximum number of todo reminders per todo item.

Default: 2

Why this value?

• Prevents infinite loops when model doesn’t update todo status
• Gives model multiple chances to notice pending work
• After limit, stops reminding (assumes model won’t update)

iteration_warning_threshold: usize

Iteration count threshold for warning user.

Default: max_iterations - 10 (190)

Why this value?

• Warns user 10 iterations before limit
• Gives time to decide: continue, break down task, or cancel
• Prevents surprising cutoff at exactly max_iterations

compression: CompressionConfig

Context compression configuration.

Controls when and how to compress long conversations. See CompressionConfig for details.

max_tokens: u32

Maximum tokens for API responses.

Default: 4096 (from QUICK_ACTION_MAX_TOKENS)

context_size_override: Option<u32>

Override for context window size.

Default: None (use provider’s default)

think: bool

Enable thinking mode for extended reasoning.

Default: false

verify_strategy: VerificationStrategy

Code verification strategy for write operations.

Controls whether and when code quality checks run on edit/write/multi_edit.

• None: No verification
• Post: Verify after write (default)
• Pre: Verify before write, block if errors
• PreQuick: Quick syntax check before, full check after

project_path: Option<PathBuf>

Project root path for code verification.

Used to detect project type (Rust/Node.js/Python/Go) and run appropriate verification commands (cargo check, tsc, etc.).

Implementations

impl AgentConfig

pub fn new( max_tokens: u32, context_size_override: Option<u32>, think: bool, compression: CompressionConfig, ) -> Self

Create a new config with all settings.

pub fn with_verify_strategy( verify_strategy: VerificationStrategy, project_path: Option<PathBuf>, ) -> Self

Create a new config with verification strategy.

pub fn with_max_iterations(max_iterations: usize) -> Self

Create a new config with custom max_iterations.

Automatically adjusts iteration_warning_threshold to max_iterations - 10.

pub fn with_compression(compression: CompressionConfig) -> Self

Create a new config with custom compression settings.

pub fn simple_task() -> Self

Create a config optimized for simple tasks.

Use for: file edits, quick queries, simple tool calls. Characteristics: lower iteration limit, smaller result size.

pub fn complex_task() -> Self

Create a config optimized for complex tasks.

Use for: refactoring, build systems, multi-file changes. Characteristics: higher iteration limit, larger result size.

pub fn max_tokens(&self) -> u32

Get max tokens for API responses.

pub fn context_size_override(&self) -> Option<u32>

Get context size override.

pub fn think(&self) -> bool

Get thinking mode flag.

pub fn compression_config(&self) -> &CompressionConfig

Get compression config.

pub fn compression_config_mut(&mut self) -> &mut CompressionConfig

Get mutable compression config.

pub fn verify_strategy(&self) -> VerificationStrategy

Get verification strategy.

pub fn project_path(&self) -> Option<&Path>

Get project path for verification.

Trait Implementations

impl Clone for AgentConfig

fn clone(&self) -> AgentConfig

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for AgentConfig

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

Formats the value using the given formatter.

impl Default for AgentConfig

fn default() -> Self

Returns the “default value” for a type.