Struct ResolvedRuntime 

pub struct ResolvedRuntime {

    pub compact_context: bool,
    pub max_tool_iterations: usize,
    pub max_execution_tree_iterations: Option<usize>,
    pub max_history_messages: usize,
    pub max_context_tokens: Option<usize>,
    pub model_context_window: usize,
    pub model_context_window_source: ModelContextWindowSource,
    pub context_compact_ratio: Option<f64>,
    pub parallel_tools: bool,
    pub tool_dispatcher: String,
    pub strict_tool_parsing: bool,
    pub tool_call_dedup_exempt: Vec<String>,
    pub tool_filter_groups: Vec<ToolFilterGroup>,
    pub max_system_prompt_chars: usize,
    pub thinking: ThinkingConfig,
    pub history_pruning: HistoryPrunerConfig,
    pub eval: EvalConfig,
    pub auto_classify: Option<AutoClassifyConfig>,
    pub effort_routing: Option<EffortRoutingConfig>,
    pub context_compression: ContextCompressionConfig,
    pub max_tool_result_chars: usize,
    pub keep_tool_context_turns: usize,
    pub tool_receipts: ToolReceiptsConfig,
    pub prompt_injection_mode: SkillsPromptInjectionMode,
}

Runtime tunables resolved from the agent’s runtime profile. Populated by Config::resolved_agent_config; never deserialized from the agent table. The runtime profile is the sole config surface for these.

Fields

compact_context: bool

max_tool_iterations: usize

max_execution_tree_iterations: Option<usize>

max_history_messages: usize

History retention limit. Structured Agent and legacy loop sessions interpret this as complete turns; channel caches retain message rows.

max_context_tokens: Option<usize>

Optional operator context budget from runtime_profiles.<name>.max_context_tokens. Without an opt-in ratio, None preserves the legacy 32,000-token budget. With a ratio, Some(n) clamps the model-relative threshold down to n. Every positive result is also capped by the selected model capacity. Some(0) disables proactive token-budget trimming. NOT the provider max_tokens output limit.

model_context_window: usize

Model’s context window (max input tokens) — from provider config.

model_context_window_source: ModelContextWindowSource

Whether model_context_window came from the selected provider profile or from the compatibility fallback for unknown/unconfigured capacity.

context_compact_ratio: Option<f64>

Opt-in fraction of model_context_window at which proactive trimming triggers. None preserves the legacy absolute-budget behavior.

parallel_tools: bool

tool_dispatcher: String

strict_tool_parsing: bool

tool_call_dedup_exempt: Vec<String>

tool_filter_groups: Vec<ToolFilterGroup>

max_system_prompt_chars: usize

thinking: ThinkingConfig

history_pruning: HistoryPrunerConfig

eval: EvalConfig

auto_classify: Option<AutoClassifyConfig>

effort_routing: Option<EffortRoutingConfig>

context_compression: ContextCompressionConfig

max_tool_result_chars: usize

keep_tool_context_turns: usize

tool_receipts: ToolReceiptsConfig

prompt_injection_mode: SkillsPromptInjectionMode

Implementations

impl ResolvedRuntime

pub fn context_limits(&self) -> ResolvedContextLimits

Resolve capacity and proactive budget for this runtime snapshot.

pub fn context_limits_for_model_window( &self, model_context_window: usize, ) -> ResolvedContextLimits

Apply this runtime policy to a route-selected model capacity.

pub fn effective_context_budget(&self) -> usize

Effective token budget for preemptive whole-turn history trimming.

Trait Implementations

impl Clone for ResolvedRuntime

fn clone(&self) -> ResolvedRuntime

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for ResolvedRuntime

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

Formats the value using the given formatter.

impl Default for ResolvedRuntime

fn default() -> ResolvedRuntime

Returns the “default value” for a type.