Struct LoopConfigSnapshot 

pub struct LoopConfigSnapshot {
    pub model: String,
    pub provider: String,
    pub config_id: Option<String>,
    pub name: Option<String>,
    pub api: Option<ApiProtocol>,
    pub base_url: Option<String>,
    pub reasoning: Option<bool>,
    pub context_window: Option<u32>,
    pub max_tokens: Option<u32>,
    pub thinking_level: Option<ThinkingLevel>,
    pub temperature: Option<f32>,
}

A lightweight, serialisable snapshot of the model that ran a loop.

Why not store the full AgentLoopConfig?

AgentLoopConfig contains API keys (in ModelConfig.api_key) and non-serialisable hook closures (BeforeTurnFn, AfterTurnFn, etc.). Storing the full config would require stripping secrets and skipping closures, yielding little extra value.

LoopConfigSnapshot captures just enough to:

• Identify which model/provider produced the messages (cost attribution, analysis).
• Support replay by telling the caller which config to reconstruct.
• Distinguish branches in evaluational parallelism (e.g. “haiku vs. opus”).
• Track per-loop config (thinking_level, temperature) for debugging.

Populated from AgentStart.config_snapshot (preferred) or extracted from the first Message::Assistant seen in the loop (fallback for older sessions).

New fields (added after the initial struct) are Option with #[serde(default, skip_serializing_if = "Option::is_none")] for backward compatibility with existing serialized sessions.

Fields

model: String

The model id string (e.g. "claude-opus-4-6", "gpt-4o").

provider: String

Provider name (e.g. "anthropic", "openai").

config_id: Option<String>

The stable config identity from AgentLoopConfig.config_id (if set).

Matches the config_segment component embedded in the loop_id format {session_id}.{config_segment}.{N}. Useful to correlate a LoopRecord back to its named configuration.

name: Option<String>

Human-friendly model name (e.g. "Claude Sonnet 4", "GPT-4o").

api: Option<ApiProtocol>

Which API protocol was used.

base_url: Option<String>

Base URL for API requests (useful for debugging which endpoint was hit).

reasoning: Option<bool>

Whether this model supports reasoning/thinking.

context_window: Option<u32>

Context window size in tokens.

max_tokens: Option<u32>

Default max output tokens.

thinking_level: Option<ThinkingLevel>

Thinking/reasoning level used for this loop.

temperature: Option<f32>

Sampling temperature used for this loop.

Trait Implementations

impl Clone for LoopConfigSnapshot

fn clone(&self) -> LoopConfigSnapshot

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for LoopConfigSnapshot

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for LoopConfigSnapshot

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for LoopConfigSnapshot

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

where __S: Serializer,

Serialize this value into the given Serde serializer.