refactor: rename pod crate to worker

crates/llm-engine/README.md

@@ -16,10 +16,10 @@ Owns:
 
 Does not own:
 
-- Pod names, sockets, process lifecycle, or scope delegation (`pod`)
+- Worker names, sockets, process lifecycle, or scope delegation (`worker`)
 - product CLI shape (`yoi`)
 - provider catalog and secret resolution (`provider`, `secrets`)
-- durable Pod current state (`pod-store`)
+- durable Worker current state (`pod-store`)
 
 ## Design notes
 

crates/llm-engine/src/engine.rs

@@ -213,7 +213,7 @@ pub struct Engine<C: LlmClient, S: EngineState = Mutable> {
     /// stream events become visible.
     lifecycle_trace_cbs: Vec<Arc<dyn Fn(usize, usize, &str, &Value) + Send + Sync>>,
     /// Non-fatal warning callbacks. Invoked when the Engine wants to
-    /// surface an advisory message to the upper layer (e.g. Pod) so it
+    /// surface an advisory message to the upper layer (e.g. Worker) so it
     /// can be forwarded to the user — distinct from `tracing::warn!`,
     /// which is for developer-facing logs.
     warning_cbs: Vec<Box<dyn Fn(&str) + Send + Sync>>,
@@ -253,7 +253,7 @@ pub struct Engine<C: LlmClient, S: EngineState = Mutable> {
     /// Plumbed into [`Request::cache_anchor`] at request build time.
     cache_anchor: Option<usize>,
     /// Conversation-scoped cache key, set by higher layers. Plumbed into
-    /// [`Request::cache_key`] at request build time. Pod 側では
+    /// [`Request::cache_key`] at request build time. Worker 側では
     /// `SegmentId` を渡す。
     cache_key: Option<String>,
     /// State marker
@@ -487,7 +487,7 @@ impl<C: LlmClient, S: EngineState> Engine<C, S> {
     /// Fired after `post_tool_call` interceptors and any `content`
     /// truncation from `tool_output_limits`, so the callback observes
     /// exactly what is persisted to history. Intended for upper layers
-    /// (e.g. Pod) to forward tool results to clients.
+    /// (e.g. Worker) to forward tool results to clients.
     pub fn on_tool_result(&mut self, callback: impl Fn(&ToolResult) + Send + Sync + 'static) {
         self.tool_result_cbs.push(Box::new(callback));
     }
@@ -1121,7 +1121,7 @@ impl<C: LlmClient, S: EngineState> Engine<C, S> {
             }
 
             // Drain interceptor-side inputs that are meant to land in
-            // history (notifications, cross-Pod events, system
+            // history (notifications, cross-Worker events, system
             // reminders). These are committed *before* the per-request
             // clone so they participate in the LLM request below and
             // get persisted by the upper layer that owns history.json.
@@ -1302,7 +1302,7 @@ impl<C: LlmClient, S: EngineState> Engine<C, S> {
 
             // Collect and commit assistant items. Routed through
             // `append_history_items` so observers (e.g. the
-            // Pod-side per-item session-log committer) see each item
+            // Worker-side per-item session-log committer) see each item
             // as it lands.
             let reasoning_items = self.thinking_block_collector.take_collected();
             let text_blocks = self.text_block_collector.take_collected();
@@ -1603,7 +1603,7 @@ impl<C: LlmClient, S: EngineState> Engine<C, S> {
             }
             Ok(ToolExecutionResult::Completed(results)) => {
                 // Route per-result pushes through the callback path so
-                // observers (e.g. the Pod-side per-item session-log
+                // observers (e.g. the Worker-side per-item session-log
                 // committer) see each tool result as it lands.
                 let items = results.into_iter().map(|result| {
                     Item::tool_result_item(
@@ -1708,7 +1708,7 @@ impl<C: LlmClient> Engine<C, Mutable> {
     /// Install byte-size caps for tool execution `content`.
     ///
     /// Passing `None` (the default) disables truncation. Higher layers
-    /// (e.g. Pod) translate manifest configuration into a concrete
+    /// (e.g. Worker) translate manifest configuration into a concrete
     /// [`ToolOutputLimits`] and install it here.
     pub fn set_tool_output_limits(&mut self, limits: Option<ToolOutputLimits>) {
         self.tool_output_limits = limits;

crates/llm-engine/src/interceptor.rs

@@ -1,6 +1,6 @@
 //! Interceptor - control flow delegation for the Engine execution loop
 //!
-//! Defines the [`Interceptor`] trait that upper layers (e.g. Pod) implement
+//! Defines the [`Interceptor`] trait that upper layers (e.g. Worker) implement
 //! to inject orchestration decisions (approval, skip, pause, abort)
 //! into the Engine's turn loop without the Engine knowing about
 //! higher-level concepts.
@@ -132,7 +132,7 @@ pub struct ToolResultInfo {
 /// Intercepts the Engine execution loop at key decision points.
 ///
 /// All methods have default implementations that let the Engine
-/// proceed without intervention. Upper layers (e.g. Pod) provide
+/// proceed without intervention. Upper layers (e.g. Worker) provide
 /// richer implementations for approval flows, permission checks, etc.
 #[async_trait]
 pub trait Interceptor: Send + Sync {
@@ -149,7 +149,7 @@ pub trait Interceptor: Send + Sync {
     ///
     /// Use this for inputs that arrive from outside the LLM and need
     /// to be reflected in the on-disk history — notifications,
-    /// cross-Pod events, system reminders. Do **not** use
+    /// cross-Worker events, system reminders. Do **not** use
     /// [`Self::pre_llm_request`] for that purpose: it mutates a
     /// per-request clone, so any committed assistant response that
     /// reacts to the injection would have no visible trigger on the

crates/llm-engine/src/llm_client/scheme/openai_responses/request.rs

@@ -51,7 +51,7 @@ pub(crate) struct ResponsesRequest {
     #[serde(skip_serializing_if = "Option::is_none")]
     pub top_p: Option<f32>,
     /// 会話単位の安定キー。ChatGPT backend (codex-oauth) は明示キーが
-    /// 無いとプロンプトキャッシュがほぼ効かない。pod 側は `SegmentId`
+    /// 無いとプロンプトキャッシュがほぼ効かない。worker 側は `SegmentId`
     /// を渡す。`Request::cache_key` が `None` のときはキー自体を送らない。
     #[serde(skip_serializing_if = "Option::is_none")]
     pub prompt_cache_key: Option<String>,

crates/llm-engine/src/llm_client/types.rs

@@ -523,7 +523,7 @@ pub struct Request {
     /// 会話単位の安定キー。`prompt_cache_key` として送られる
     /// (OpenAI Responses)。ChatGPT backend (codex-oauth) は明示キーが
     /// 無いと org/project ハッシュ衝突でプロンプトキャッシュが
-    /// ほぼヒットしないため、pod 側で `SegmentId` を渡す運用を想定。
+    /// ほぼヒットしないため、worker 側で `SegmentId` を渡す運用を想定。
     /// `cache_anchor` と違い名前空間キーであり、`prefix anchor` とは
     /// 別の概念。`cache_anchor` を読まない provider と同じく、
     /// `prompt_cache_key` を持たない provider は無視する。

crates/llm-engine/src/prune.rs

@@ -8,7 +8,7 @@
 //!
 //! Prune は **コンテキスト射影** であり、history の変換ではない。
 //! この crate が提供するのは pure な候補抽出 [`prunable_indices`] のみで、
-//! 射影の適用は上位層（`pod::prune_hook` 等）が LLM に送る一時コンテキスト
+//! 射影の適用は上位層（`worker::prune_hook` 等）が LLM に送る一時コンテキスト
 //! に対してだけ行う。Engine の永続履歴は決して変更されない。
 //!
 //! 保護境界は末尾 token budget で決めるが、この crate は usage 履歴を
@@ -75,7 +75,7 @@ pub enum PruneDecision {
 }
 
 /// Optional observer invoked after each prune evaluation, regardless of
-/// branch. Pod 等の上位層が install して metrics を発行する。
+/// branch. Worker 等の上位層が install して metrics を発行する。
 pub type PruneObserver = Box<dyn Fn(&PruneEvaluation) + Send + Sync>;
 
 /// Configuration for the Prune algorithm.

crates/llm-engine/src/timeline/tool_call_collector.rs

@@ -130,13 +130,13 @@ mod tests {
         let mut timeline = Timeline::new();
         timeline.on_tool_use_block(collector.clone());
 
-        timeline.dispatch(&Event::tool_use_start(0, "tool_empty", "ListPods"));
+        timeline.dispatch(&Event::tool_use_start(0, "tool_empty", "ListWorkers"));
         timeline.dispatch(&Event::tool_use_stop(0));
 
         let calls = collector.take_collected();
         assert_eq!(calls.len(), 1);
         assert_eq!(calls[0].id, "tool_empty");
-        assert_eq!(calls[0].name, "ListPods");
+        assert_eq!(calls[0].name, "ListWorkers");
         assert!(calls[0].input.is_object());
         assert_eq!(
             calls[0].input,

crates/llm-engine/src/tool_server.rs

@@ -75,7 +75,7 @@ impl ToolServerHandle {
     /// Execute all pending factories and register the resulting tools.
     ///
     /// Called implicitly by `Engine::lock()` before the first turn.
-    /// Exposed as `pub` so higher layers (e.g. Pod) can force-materialise
+    /// Exposed as `pub` so higher layers (e.g. Worker) can force-materialise
     /// tools earlier — for example when building a system-prompt template
     /// context that needs the list of registered tool names. Redundant
     /// calls are no-ops.

crates/llm-engine/src/usage_record.rs

@@ -2,7 +2,7 @@
 //!
 //! 1 リクエストの送信時点での「ある history prefix 長で計測した占有量」を
 //! 1 件分にまとめたもの。`UsageEvent` (provider stream イベント) を
-//! 受けて呼び出し側 (typically Pod) が組み立て、永続化層
+//! 受けて呼び出し側 (typically Worker) が組み立て、永続化層
 //! (session-store) に流したり、token accounting (`token_counter`) で
 //! 履歴として参照したりする。