wip: grok provider abstraction

rust/crates/api/src/error.rs

@@ -4,7 +4,10 @@ use std::time::Duration;
 
 #[derive(Debug)]
 pub enum ApiError {
-    MissingApiKey,
+    MissingCredentials {
+        provider: &'static str,
+        env_vars: &'static [&'static str],
+    },
     ExpiredOAuthToken,
     Auth(String),
     InvalidApiKeyEnv(VarError),
@@ -30,13 +33,21 @@ pub enum ApiError {
 }
 
 impl ApiError {
+    #[must_use]
+    pub const fn missing_credentials(
+        provider: &'static str,
+        env_vars: &'static [&'static str],
+    ) -> Self {
+        Self::MissingCredentials { provider, env_vars }
+    }
+
     #[must_use]
     pub fn is_retryable(&self) -> bool {
         match self {
             Self::Http(error) => error.is_connect() || error.is_timeout() || error.is_request(),
             Self::Api { retryable, .. } => *retryable,
             Self::RetriesExhausted { last_error, .. } => last_error.is_retryable(),
-            Self::MissingApiKey
+            Self::MissingCredentials { .. }
             | Self::ExpiredOAuthToken
             | Self::Auth(_)
             | Self::InvalidApiKeyEnv(_)
@@ -51,12 +62,11 @@ impl ApiError {
 impl Display for ApiError {
     fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
         match self {
-            Self::MissingApiKey => {
-                write!(
-                    f,
-                    "ANTHROPIC_AUTH_TOKEN or ANTHROPIC_API_KEY is not set; export one before calling the Anthropic API"
-                )
-            }
+            Self::MissingCredentials { provider, env_vars } => write!(
+                f,
+                "missing {provider} credentials; export {} before calling the {provider} API",
+                env_vars.join(" or ")
+            ),
             Self::ExpiredOAuthToken => {
                 write!(
                     f,
@@ -65,10 +75,7 @@ impl Display for ApiError {
             }
             Self::Auth(message) => write!(f, "auth error: {message}"),
             Self::InvalidApiKeyEnv(error) => {
-                write!(
-                    f,
-                    "failed to read ANTHROPIC_AUTH_TOKEN / ANTHROPIC_API_KEY: {error}"
-                )
+                write!(f, "failed to read credential environment variable: {error}")
             }
             Self::Http(error) => write!(f, "http error: {error}"),
             Self::Io(error) => write!(f, "io error: {error}"),
@@ -81,20 +88,14 @@ impl Display for ApiError {
                 ..
             } => match (error_type, message) {
                 (Some(error_type), Some(message)) => {
-                    write!(
-                        f,
-                        "anthropic api returned {status} ({error_type}): {message}"
-                    )
+                    write!(f, "api returned {status} ({error_type}): {message}")
                 }
-                _ => write!(f, "anthropic api returned {status}: {body}"),
+                _ => write!(f, "api returned {status}: {body}"),
             },
             Self::RetriesExhausted {
                 attempts,
                 last_error,
-            } => write!(
-                f,
-                "anthropic api failed after {attempts} attempts: {last_error}"
-            ),
+            } => write!(f, "api failed after {attempts} attempts: {last_error}"),
             Self::InvalidSseFrame(message) => write!(f, "invalid sse frame: {message}"),
             Self::BackoffOverflow {
                 attempt,

rust/crates/api/src/lib.rs

@@ -1,13 +1,19 @@
 mod client;
 mod error;
+mod providers;
 mod sse;
 mod types;
 
 pub use client::{
-    oauth_token_is_expired, read_base_url, resolve_saved_oauth_token, resolve_startup_auth_source,
-    AnthropicClient, AuthSource, MessageStream, OAuthTokenSet,
+    oauth_token_is_expired, read_base_url, read_xai_base_url, resolve_saved_oauth_token,
+    resolve_startup_auth_source, MessageStream, OAuthTokenSet, ProviderClient,
 };
 pub use error::ApiError;
+pub use providers::anthropic::{AnthropicClient, AuthSource};
+pub use providers::openai_compat::{OpenAiCompatClient, OpenAiCompatConfig};
+pub use providers::{
+    detect_provider_kind, max_tokens_for_model, resolve_model_alias, ProviderKind,
+};
 pub use sse::{parse_frame, SseParser};
 pub use types::{
     ContentBlockDelta, ContentBlockDeltaEvent, ContentBlockStartEvent, ContentBlockStopEvent,

rust/crates/api/src/providers/mod.rs

@@ -0,0 +1,216 @@
+use std::future::Future;
+use std::pin::Pin;
+
+use crate::error::ApiError;
+use crate::types::{MessageRequest, MessageResponse};
+
+pub mod anthropic;
+pub mod openai_compat;
+
+pub type ProviderFuture<'a, T> = Pin<Box<dyn Future<Output = Result<T, ApiError>> + Send + 'a>>;
+
+pub trait Provider {
+    type Stream;
+
+    fn send_message<'a>(
+        &'a self,
+        request: &'a MessageRequest,
+    ) -> ProviderFuture<'a, MessageResponse>;
+
+    fn stream_message<'a>(
+        &'a self,
+        request: &'a MessageRequest,
+    ) -> ProviderFuture<'a, Self::Stream>;
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub enum ProviderKind {
+    Anthropic,
+    Xai,
+    OpenAi,
+}
+
+#[derive(Debug, Clone, Copy, PartialEq, Eq)]
+pub struct ProviderMetadata {
+    pub provider: ProviderKind,
+    pub auth_env: &'static str,
+    pub base_url_env: &'static str,
+    pub default_base_url: &'static str,
+}
+
+const MODEL_REGISTRY: &[(&str, ProviderMetadata)] = &[
+    (
+        "opus",
+        ProviderMetadata {
+            provider: ProviderKind::Anthropic,
+            auth_env: "ANTHROPIC_API_KEY",
+            base_url_env: "ANTHROPIC_BASE_URL",
+            default_base_url: anthropic::DEFAULT_BASE_URL,
+        },
+    ),
+    (
+        "sonnet",
+        ProviderMetadata {
+            provider: ProviderKind::Anthropic,
+            auth_env: "ANTHROPIC_API_KEY",
+            base_url_env: "ANTHROPIC_BASE_URL",
+            default_base_url: anthropic::DEFAULT_BASE_URL,
+        },
+    ),
+    (
+        "haiku",
+        ProviderMetadata {
+            provider: ProviderKind::Anthropic,
+            auth_env: "ANTHROPIC_API_KEY",
+            base_url_env: "ANTHROPIC_BASE_URL",
+            default_base_url: anthropic::DEFAULT_BASE_URL,
+        },
+    ),
+    (
+        "grok",
+        ProviderMetadata {
+            provider: ProviderKind::Xai,
+            auth_env: "XAI_API_KEY",
+            base_url_env: "XAI_BASE_URL",
+            default_base_url: openai_compat::DEFAULT_XAI_BASE_URL,
+        },
+    ),
+    (
+        "grok-3",
+        ProviderMetadata {
+            provider: ProviderKind::Xai,
+            auth_env: "XAI_API_KEY",
+            base_url_env: "XAI_BASE_URL",
+            default_base_url: openai_compat::DEFAULT_XAI_BASE_URL,
+        },
+    ),
+    (
+        "grok-mini",
+        ProviderMetadata {
+            provider: ProviderKind::Xai,
+            auth_env: "XAI_API_KEY",
+            base_url_env: "XAI_BASE_URL",
+            default_base_url: openai_compat::DEFAULT_XAI_BASE_URL,
+        },
+    ),
+    (
+        "grok-3-mini",
+        ProviderMetadata {
+            provider: ProviderKind::Xai,
+            auth_env: "XAI_API_KEY",
+            base_url_env: "XAI_BASE_URL",
+            default_base_url: openai_compat::DEFAULT_XAI_BASE_URL,
+        },
+    ),
+    (
+        "grok-2",
+        ProviderMetadata {
+            provider: ProviderKind::Xai,
+            auth_env: "XAI_API_KEY",
+            base_url_env: "XAI_BASE_URL",
+            default_base_url: openai_compat::DEFAULT_XAI_BASE_URL,
+        },
+    ),
+];
+
+#[must_use]
+pub fn resolve_model_alias(model: &str) -> String {
+    let trimmed = model.trim();
+    let lower = trimmed.to_ascii_lowercase();
+    MODEL_REGISTRY
+        .iter()
+        .find_map(|(alias, metadata)| {
+            (*alias == lower).then_some(match metadata.provider {
+                ProviderKind::Anthropic => match *alias {
+                    "opus" => "claude-opus-4-6",
+                    "sonnet" => "claude-sonnet-4-6",
+                    "haiku" => "claude-haiku-4-5-20251213",
+                    _ => trimmed,
+                },
+                ProviderKind::Xai => match *alias {
+                    "grok" | "grok-3" => "grok-3",
+                    "grok-mini" | "grok-3-mini" => "grok-3-mini",
+                    "grok-2" => "grok-2",
+                    _ => trimmed,
+                },
+                ProviderKind::OpenAi => trimmed,
+            })
+        })
+        .map_or_else(|| trimmed.to_string(), ToOwned::to_owned)
+}
+
+#[must_use]
+pub fn metadata_for_model(model: &str) -> Option<ProviderMetadata> {
+    let canonical = resolve_model_alias(model);
+    if canonical.starts_with("claude") {
+        return Some(ProviderMetadata {
+            provider: ProviderKind::Anthropic,
+            auth_env: "ANTHROPIC_API_KEY",
+            base_url_env: "ANTHROPIC_BASE_URL",
+            default_base_url: anthropic::DEFAULT_BASE_URL,
+        });
+    }
+    if canonical.starts_with("grok") {
+        return Some(ProviderMetadata {
+            provider: ProviderKind::Xai,
+            auth_env: "XAI_API_KEY",
+            base_url_env: "XAI_BASE_URL",
+            default_base_url: openai_compat::DEFAULT_XAI_BASE_URL,
+        });
+    }
+    None
+}
+
+#[must_use]
+pub fn detect_provider_kind(model: &str) -> ProviderKind {
+    if let Some(metadata) = metadata_for_model(model) {
+        return metadata.provider;
+    }
+    if anthropic::has_auth_from_env_or_saved().unwrap_or(false) {
+        return ProviderKind::Anthropic;
+    }
+    if openai_compat::has_api_key("OPENAI_API_KEY") {
+        return ProviderKind::OpenAi;
+    }
+    if openai_compat::has_api_key("XAI_API_KEY") {
+        return ProviderKind::Xai;
+    }
+    ProviderKind::Anthropic
+}
+
+#[must_use]
+pub fn max_tokens_for_model(model: &str) -> u32 {
+    let canonical = resolve_model_alias(model);
+    if canonical.contains("opus") {
+        32_000
+    } else {
+        64_000
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::{detect_provider_kind, max_tokens_for_model, resolve_model_alias, ProviderKind};
+
+    #[test]
+    fn resolves_grok_aliases() {
+        assert_eq!(resolve_model_alias("grok"), "grok-3");
+        assert_eq!(resolve_model_alias("grok-mini"), "grok-3-mini");
+        assert_eq!(resolve_model_alias("grok-2"), "grok-2");
+    }
+
+    #[test]
+    fn detects_provider_from_model_name_first() {
+        assert_eq!(detect_provider_kind("grok"), ProviderKind::Xai);
+        assert_eq!(
+            detect_provider_kind("claude-sonnet-4-6"),
+            ProviderKind::Anthropic
+        );
+    }
+
+    #[test]
+    fn keeps_existing_max_token_heuristic() {
+        assert_eq!(max_tokens_for_model("opus"), 32_000);
+        assert_eq!(max_tokens_for_model("grok-3"), 64_000);
+    }
+}

rust/crates/api/tests/client_integration.rs

@@ -3,9 +3,9 @@ use std::sync::Arc;
 use std::time::Duration;
 
 use api::{
-    AnthropicClient, ApiError, ContentBlockDelta, ContentBlockDeltaEvent, ContentBlockStartEvent,
-    InputContentBlock, InputMessage, MessageDeltaEvent, MessageRequest, OutputContentBlock,
-    StreamEvent, ToolChoice, ToolDefinition,
+    AnthropicClient, ApiError, AuthSource, ContentBlockDelta, ContentBlockDeltaEvent,
+    ContentBlockStartEvent, InputContentBlock, InputMessage, MessageDeltaEvent, MessageRequest,
+    OutputContentBlock, ProviderClient, StreamEvent, ToolChoice, ToolDefinition,
 };
 use serde_json::json;
 use tokio::io::{AsyncReadExt, AsyncWriteExt};
@@ -195,6 +195,47 @@ async fn retries_retryable_failures_before_succeeding() {
     assert_eq!(state.lock().await.len(), 2);
 }
 
+#[tokio::test]
+async fn provider_client_dispatches_anthropic_requests() {
+    let state = Arc::new(Mutex::new(Vec::<CapturedRequest>::new()));
+    let server = spawn_server(
+        state.clone(),
+        vec![http_response(
+            "200 OK",
+            "application/json",
+            "{\"id\":\"msg_provider\",\"type\":\"message\",\"role\":\"assistant\",\"content\":[{\"type\":\"text\",\"text\":\"Dispatched\"}],\"model\":\"claude-3-7-sonnet-latest\",\"stop_reason\":\"end_turn\",\"stop_sequence\":null,\"usage\":{\"input_tokens\":3,\"output_tokens\":2}}",
+        )],
+    )
+    .await;
+
+    let client = ProviderClient::from_model_with_anthropic_auth(
+        "claude-sonnet-4-6",
+        Some(AuthSource::ApiKey("test-key".to_string())),
+    )
+    .expect("anthropic provider client should be constructed");
+    let client = match client {
+        ProviderClient::Anthropic(client) => {
+            ProviderClient::Anthropic(client.with_base_url(server.base_url()))
+        }
+        other => panic!("expected anthropic provider, got {other:?}"),
+    };
+
+    let response = client
+        .send_message(&sample_request(false))
+        .await
+        .expect("provider-dispatched request should succeed");
+
+    assert_eq!(response.total_tokens(), 5);
+
+    let captured = state.lock().await;
+    let request = captured.first().expect("server should capture request");
+    assert_eq!(request.path, "/v1/messages");
+    assert_eq!(
+        request.headers.get("x-api-key").map(String::as_str),
+        Some("test-key")
+    );
+}
+
 #[tokio::test]
 async fn surfaces_retry_exhaustion_for_persistent_retryable_errors() {
     let state = Arc::new(Mutex::new(Vec::<CapturedRequest>::new()));

rust/crates/api/tests/openai_compat_integration.rs

@@ -0,0 +1,415 @@
+use std::collections::HashMap;
+use std::ffi::OsString;
+use std::sync::Arc;
+use std::sync::{Mutex as StdMutex, OnceLock};
+
+use api::{
+    ContentBlockDelta, ContentBlockDeltaEvent, ContentBlockStartEvent, ContentBlockStopEvent,
+    InputContentBlock, InputMessage, MessageRequest, OpenAiCompatClient, OpenAiCompatConfig,
+    OutputContentBlock, ProviderClient, StreamEvent, ToolChoice, ToolDefinition,
+};
+use serde_json::json;
+use tokio::io::{AsyncReadExt, AsyncWriteExt};
+use tokio::net::TcpListener;
+use tokio::sync::Mutex;
+
+#[tokio::test]
+async fn send_message_uses_openai_compatible_endpoint_and_auth() {
+    let state = Arc::new(Mutex::new(Vec::<CapturedRequest>::new()));
+    let body = concat!(
+        "{",
+        "\"id\":\"chatcmpl_test\",",
+        "\"model\":\"grok-3\",",
+        "\"choices\":[{",
+        "\"message\":{\"role\":\"assistant\",\"content\":\"Hello from Grok\",\"tool_calls\":[]},",
+        "\"finish_reason\":\"stop\"",
+        "}],",
+        "\"usage\":{\"prompt_tokens\":11,\"completion_tokens\":5}",
+        "}"
+    );
+    let server = spawn_server(
+        state.clone(),
+        vec![http_response("200 OK", "application/json", body)],
+    )
+    .await;
+
+    let client = OpenAiCompatClient::new("xai-test-key", OpenAiCompatConfig::xai())
+        .with_base_url(server.base_url());
+    let response = client
+        .send_message(&sample_request(false))
+        .await
+        .expect("request should succeed");
+
+    assert_eq!(response.model, "grok-3");
+    assert_eq!(response.total_tokens(), 16);
+    assert_eq!(
+        response.content,
+        vec![OutputContentBlock::Text {
+            text: "Hello from Grok".to_string(),
+        }]
+    );
+
+    let captured = state.lock().await;
+    let request = captured.first().expect("server should capture request");
+    assert_eq!(request.path, "/chat/completions");
+    assert_eq!(
+        request.headers.get("authorization").map(String::as_str),
+        Some("Bearer xai-test-key")
+    );
+    let body: serde_json::Value = serde_json::from_str(&request.body).expect("json body");
+    assert_eq!(body["model"], json!("grok-3"));
+    assert_eq!(body["messages"][0]["role"], json!("system"));
+    assert_eq!(body["tools"][0]["type"], json!("function"));
+}
+
+#[tokio::test]
+async fn send_message_accepts_full_chat_completions_endpoint_override() {
+    let state = Arc::new(Mutex::new(Vec::<CapturedRequest>::new()));
+    let body = concat!(
+        "{",
+        "\"id\":\"chatcmpl_full_endpoint\",",
+        "\"model\":\"grok-3\",",
+        "\"choices\":[{",
+        "\"message\":{\"role\":\"assistant\",\"content\":\"Endpoint override works\",\"tool_calls\":[]},",
+        "\"finish_reason\":\"stop\"",
+        "}],",
+        "\"usage\":{\"prompt_tokens\":7,\"completion_tokens\":3}",
+        "}"
+    );
+    let server = spawn_server(
+        state.clone(),
+        vec![http_response("200 OK", "application/json", body)],
+    )
+    .await;
+
+    let endpoint_url = format!("{}/chat/completions", server.base_url());
+    let client = OpenAiCompatClient::new("xai-test-key", OpenAiCompatConfig::xai())
+        .with_base_url(endpoint_url);
+    let response = client
+        .send_message(&sample_request(false))
+        .await
+        .expect("request should succeed");
+
+    assert_eq!(response.total_tokens(), 10);
+
+    let captured = state.lock().await;
+    let request = captured.first().expect("server should capture request");
+    assert_eq!(request.path, "/chat/completions");
+}
+
+#[tokio::test]
+async fn stream_message_normalizes_text_and_multiple_tool_calls() {
+    let state = Arc::new(Mutex::new(Vec::<CapturedRequest>::new()));
+    let sse = concat!(
+        "data: {\"id\":\"chatcmpl_stream\",\"model\":\"grok-3\",\"choices\":[{\"delta\":{\"content\":\"Hello\"}}]}\n\n",
+        "data: {\"id\":\"chatcmpl_stream\",\"choices\":[{\"delta\":{\"tool_calls\":[{\"index\":0,\"id\":\"call_1\",\"function\":{\"name\":\"weather\",\"arguments\":\"{\\\"city\\\":\\\"Paris\\\"}\"}},{\"index\":1,\"id\":\"call_2\",\"function\":{\"name\":\"clock\",\"arguments\":\"{\\\"zone\\\":\\\"UTC\\\"}\"}}]}}]}\n\n",
+        "data: {\"id\":\"chatcmpl_stream\",\"choices\":[{\"delta\":{},\"finish_reason\":\"tool_calls\"}]}\n\n",
+        "data: [DONE]\n\n"
+    );
+    let server = spawn_server(
+        state.clone(),
+        vec![http_response_with_headers(
+            "200 OK",
+            "text/event-stream",
+            sse,
+            &[("x-request-id", "req_grok_stream")],
+        )],
+    )
+    .await;
+
+    let client = OpenAiCompatClient::new("xai-test-key", OpenAiCompatConfig::xai())
+        .with_base_url(server.base_url());
+    let mut stream = client
+        .stream_message(&sample_request(false))
+        .await
+        .expect("stream should start");
+
+    assert_eq!(stream.request_id(), Some("req_grok_stream"));
+
+    let mut events = Vec::new();
+    while let Some(event) = stream.next_event().await.expect("event should parse") {
+        events.push(event);
+    }
+
+    assert!(matches!(events[0], StreamEvent::MessageStart(_)));
+    assert!(matches!(
+        events[1],
+        StreamEvent::ContentBlockStart(ContentBlockStartEvent {
+            content_block: OutputContentBlock::Text { .. },
+            ..
+        })
+    ));
+    assert!(matches!(
+        events[2],
+        StreamEvent::ContentBlockDelta(ContentBlockDeltaEvent {
+            delta: ContentBlockDelta::TextDelta { .. },
+            ..
+        })
+    ));
+    assert!(matches!(
+        events[3],
+        StreamEvent::ContentBlockStart(ContentBlockStartEvent {
+            index: 1,
+            content_block: OutputContentBlock::ToolUse { .. },
+        })
+    ));
+    assert!(matches!(
+        events[4],
+        StreamEvent::ContentBlockDelta(ContentBlockDeltaEvent {
+            index: 1,
+            delta: ContentBlockDelta::InputJsonDelta { .. },
+        })
+    ));
+    assert!(matches!(
+        events[5],
+        StreamEvent::ContentBlockStart(ContentBlockStartEvent {
+            index: 2,
+            content_block: OutputContentBlock::ToolUse { .. },
+        })
+    ));
+    assert!(matches!(
+        events[6],
+        StreamEvent::ContentBlockDelta(ContentBlockDeltaEvent {
+            index: 2,
+            delta: ContentBlockDelta::InputJsonDelta { .. },
+        })
+    ));
+    assert!(matches!(
+        events[7],
+        StreamEvent::ContentBlockStop(ContentBlockStopEvent { index: 1 })
+    ));
+    assert!(matches!(
+        events[8],
+        StreamEvent::ContentBlockStop(ContentBlockStopEvent { index: 2 })
+    ));
+    assert!(matches!(
+        events[9],
+        StreamEvent::ContentBlockStop(ContentBlockStopEvent { index: 0 })
+    ));
+    assert!(matches!(events[10], StreamEvent::MessageDelta(_)));
+    assert!(matches!(events[11], StreamEvent::MessageStop(_)));
+
+    let captured = state.lock().await;
+    let request = captured.first().expect("captured request");
+    assert_eq!(request.path, "/chat/completions");
+    assert!(request.body.contains("\"stream\":true"));
+}
+
+#[tokio::test]
+async fn provider_client_dispatches_xai_requests_from_env() {
+    let _lock = env_lock();
+    let _api_key = ScopedEnvVar::set("XAI_API_KEY", "xai-test-key");
+
+    let state = Arc::new(Mutex::new(Vec::<CapturedRequest>::new()));
+    let server = spawn_server(
+        state.clone(),
+        vec![http_response(
+            "200 OK",
+            "application/json",
+            "{\"id\":\"chatcmpl_provider\",\"model\":\"grok-3\",\"choices\":[{\"message\":{\"role\":\"assistant\",\"content\":\"Through provider client\",\"tool_calls\":[]},\"finish_reason\":\"stop\"}],\"usage\":{\"prompt_tokens\":9,\"completion_tokens\":4}}",
+        )],
+    )
+    .await;
+    let _base_url = ScopedEnvVar::set("XAI_BASE_URL", server.base_url());
+
+    let client =
+        ProviderClient::from_model("grok").expect("xAI provider client should be constructed");
+    assert!(matches!(client, ProviderClient::Xai(_)));
+
+    let response = client
+        .send_message(&sample_request(false))
+        .await
+        .expect("provider-dispatched request should succeed");
+
+    assert_eq!(response.total_tokens(), 13);
+
+    let captured = state.lock().await;
+    let request = captured.first().expect("captured request");
+    assert_eq!(request.path, "/chat/completions");
+    assert_eq!(
+        request.headers.get("authorization").map(String::as_str),
+        Some("Bearer xai-test-key")
+    );
+}
+
+#[derive(Debug, Clone, PartialEq, Eq)]
+struct CapturedRequest {
+    path: String,
+    headers: HashMap<String, String>,
+    body: String,
+}
+
+struct TestServer {
+    base_url: String,
+    join_handle: tokio::task::JoinHandle<()>,
+}
+
+impl TestServer {
+    fn base_url(&self) -> String {
+        self.base_url.clone()
+    }
+}
+
+impl Drop for TestServer {
+    fn drop(&mut self) {
+        self.join_handle.abort();
+    }
+}
+
+async fn spawn_server(
+    state: Arc<Mutex<Vec<CapturedRequest>>>,
+    responses: Vec<String>,
+) -> TestServer {
+    let listener = TcpListener::bind("127.0.0.1:0")
+        .await
+        .expect("listener should bind");
+    let address = listener.local_addr().expect("listener addr");
+    let join_handle = tokio::spawn(async move {
+        for response in responses {
+            let (mut socket, _) = listener.accept().await.expect("accept");
+            let mut buffer = Vec::new();
+            let mut header_end = None;
+            loop {
+                let mut chunk = [0_u8; 1024];
+                let read = socket.read(&mut chunk).await.expect("read request");
+                if read == 0 {
+                    break;
+                }
+                buffer.extend_from_slice(&chunk[..read]);
+                if let Some(position) = find_header_end(&buffer) {
+                    header_end = Some(position);
+                    break;
+                }
+            }
+
+            let header_end = header_end.expect("headers should exist");
+            let (header_bytes, remaining) = buffer.split_at(header_end);
+            let header_text = String::from_utf8(header_bytes.to_vec()).expect("utf8 headers");
+            let mut lines = header_text.split("\r\n");
+            let request_line = lines.next().expect("request line");
+            let path = request_line
+                .split_whitespace()
+                .nth(1)
+                .expect("path")
+                .to_string();
+            let mut headers = HashMap::new();
+            let mut content_length = 0_usize;
+            for line in lines {
+                if line.is_empty() {
+                    continue;
+                }
+                let (name, value) = line.split_once(':').expect("header");
+                let value = value.trim().to_string();
+                if name.eq_ignore_ascii_case("content-length") {
+                    content_length = value.parse().expect("content length");
+                }
+                headers.insert(name.to_ascii_lowercase(), value);
+            }
+
+            let mut body = remaining[4..].to_vec();
+            while body.len() < content_length {
+                let mut chunk = vec![0_u8; content_length - body.len()];
+                let read = socket.read(&mut chunk).await.expect("read body");
+                if read == 0 {
+                    break;
+                }
+                body.extend_from_slice(&chunk[..read]);
+            }
+
+            state.lock().await.push(CapturedRequest {
+                path,
+                headers,
+                body: String::from_utf8(body).expect("utf8 body"),
+            });
+
+            socket
+                .write_all(response.as_bytes())
+                .await
+                .expect("write response");
+        }
+    });
+
+    TestServer {
+        base_url: format!("http://{address}"),
+        join_handle,
+    }
+}
+
+fn find_header_end(bytes: &[u8]) -> Option<usize> {
+    bytes.windows(4).position(|window| window == b"\r\n\r\n")
+}
+
+fn http_response(status: &str, content_type: &str, body: &str) -> String {
+    http_response_with_headers(status, content_type, body, &[])
+}
+
+fn http_response_with_headers(
+    status: &str,
+    content_type: &str,
+    body: &str,
+    headers: &[(&str, &str)],
+) -> String {
+    let mut extra_headers = String::new();
+    for (name, value) in headers {
+        use std::fmt::Write as _;
+        write!(&mut extra_headers, "{name}: {value}\r\n").expect("header write");
+    }
+    format!(
+        "HTTP/1.1 {status}\r\ncontent-type: {content_type}\r\n{extra_headers}content-length: {}\r\nconnection: close\r\n\r\n{body}",
+        body.len()
+    )
+}
+
+fn sample_request(stream: bool) -> MessageRequest {
+    MessageRequest {
+        model: "grok-3".to_string(),
+        max_tokens: 64,
+        messages: vec![InputMessage {
+            role: "user".to_string(),
+            content: vec![InputContentBlock::Text {
+                text: "Say hello".to_string(),
+            }],
+        }],
+        system: Some("Use tools when needed".to_string()),
+        tools: Some(vec![ToolDefinition {
+            name: "weather".to_string(),
+            description: Some("Fetches weather".to_string()),
+            input_schema: json!({
+                "type": "object",
+                "properties": {"city": {"type": "string"}},
+                "required": ["city"]
+            }),
+        }]),
+        tool_choice: Some(ToolChoice::Auto),
+        stream,
+    }
+}
+
+fn env_lock() -> std::sync::MutexGuard<'static, ()> {
+    static LOCK: OnceLock<StdMutex<()>> = OnceLock::new();
+    LOCK.get_or_init(|| StdMutex::new(()))
+        .lock()
+        .unwrap_or_else(|poisoned| poisoned.into_inner())
+}
+
+struct ScopedEnvVar {
+    key: &'static str,
+    previous: Option<OsString>,
+}
+
+impl ScopedEnvVar {
+    fn set(key: &'static str, value: impl AsRef<std::ffi::OsStr>) -> Self {
+        let previous = std::env::var_os(key);
+        std::env::set_var(key, value);
+        Self { key, previous }
+    }
+}
+
+impl Drop for ScopedEnvVar {
+    fn drop(&mut self) {
+        match &self.previous {
+            Some(value) => std::env::set_var(self.key, value),
+            None => std::env::remove_var(self.key),
+        }
+    }
+}

rust/crates/api/tests/provider_client_integration.rs

@@ -0,0 +1,86 @@
+use std::ffi::OsString;
+use std::sync::{Mutex, OnceLock};
+
+use api::{read_xai_base_url, ApiError, AuthSource, ProviderClient, ProviderKind};
+
+#[test]
+fn provider_client_routes_grok_aliases_through_xai() {
+    let _lock = env_lock();
+    let _xai_api_key = EnvVarGuard::set("XAI_API_KEY", Some("xai-test-key"));
+
+    let client = ProviderClient::from_model("grok-mini").expect("grok alias should resolve");
+
+    assert_eq!(client.provider_kind(), ProviderKind::Xai);
+}
+
+#[test]
+fn provider_client_reports_missing_xai_credentials_for_grok_models() {
+    let _lock = env_lock();
+    let _xai_api_key = EnvVarGuard::set("XAI_API_KEY", None);
+
+    let error = ProviderClient::from_model("grok-3")
+        .expect_err("grok requests without XAI_API_KEY should fail fast");
+
+    match error {
+        ApiError::MissingCredentials { provider, env_vars } => {
+            assert_eq!(provider, "xAI");
+            assert_eq!(env_vars, &["XAI_API_KEY"]);
+        }
+        other => panic!("expected missing xAI credentials, got {other:?}"),
+    }
+}
+
+#[test]
+fn provider_client_uses_explicit_anthropic_auth_without_env_lookup() {
+    let _lock = env_lock();
+    let _anthropic_api_key = EnvVarGuard::set("ANTHROPIC_API_KEY", None);
+    let _anthropic_auth_token = EnvVarGuard::set("ANTHROPIC_AUTH_TOKEN", None);
+
+    let client = ProviderClient::from_model_with_anthropic_auth(
+        "claude-sonnet-4-6",
+        Some(AuthSource::ApiKey("anthropic-test-key".to_string())),
+    )
+    .expect("explicit anthropic auth should avoid env lookup");
+
+    assert_eq!(client.provider_kind(), ProviderKind::Anthropic);
+}
+
+#[test]
+fn read_xai_base_url_prefers_env_override() {
+    let _lock = env_lock();
+    let _xai_base_url = EnvVarGuard::set("XAI_BASE_URL", Some("https://example.xai.test/v1"));
+
+    assert_eq!(read_xai_base_url(), "https://example.xai.test/v1");
+}
+
+fn env_lock() -> std::sync::MutexGuard<'static, ()> {
+    static LOCK: OnceLock<Mutex<()>> = OnceLock::new();
+    LOCK.get_or_init(|| Mutex::new(()))
+        .lock()
+        .unwrap_or_else(|poisoned| poisoned.into_inner())
+}
+
+struct EnvVarGuard {
+    key: &'static str,
+    original: Option<OsString>,
+}
+
+impl EnvVarGuard {
+    fn set(key: &'static str, value: Option<&str>) -> Self {
+        let original = std::env::var_os(key);
+        match value {
+            Some(value) => std::env::set_var(key, value),
+            None => std::env::remove_var(key),
+        }
+        Self { key, original }
+    }
+}
+
+impl Drop for EnvVarGuard {
+    fn drop(&mut self) {
+        match &self.original {
+            Some(value) => std::env::set_var(self.key, value),
+            None => std::env::remove_var(self.key),
+        }
+    }
+}

rust/crates/commands/src/lib.rs

@@ -125,8 +125,8 @@ const SLASH_COMMAND_SPECS: &[SlashCommandSpec] = &[
     },
     SlashCommandSpec {
         name: "session",
-        summary: "List, switch, or fork managed local sessions",
-        argument_hint: Some("[list|switch <session-id>|fork [branch-name]]"),
+        summary: "List or switch managed local sessions",
+        argument_hint: Some("[list|switch <session-id>]"),
         resume_supported: false,
     },
 ];
@@ -229,7 +229,7 @@ pub fn resume_supported_slash_commands() -> Vec<&'static SlashCommandSpec> {
 pub fn render_slash_command_help() -> String {
     let mut lines = vec![
         "Slash commands".to_string(),
-        "  [resume] means the command also works with --resume SESSION.jsonl".to_string(),
+        "  [resume] means the command also works with --resume SESSION.json".to_string(),
     ];
     for spec in slash_command_specs() {
         let name = match spec.argument_hint {
@@ -365,19 +365,12 @@ mod tests {
                 target: Some("abc123".to_string())
             })
         );
-        assert_eq!(
-            SlashCommand::parse("/session fork incident-review"),
-            Some(SlashCommand::Session {
-                action: Some("fork".to_string()),
-                target: Some("incident-review".to_string())
-            })
-        );
     }
 
     #[test]
     fn renders_help_from_shared_specs() {
         let help = render_slash_command_help();
-        assert!(help.contains("works with --resume SESSION.jsonl"));
+        assert!(help.contains("works with --resume SESSION.json"));
         assert!(help.contains("/help"));
         assert!(help.contains("/status"));
         assert!(help.contains("/compact"));
@@ -392,24 +385,26 @@ mod tests {
         assert!(help.contains("/diff"));
         assert!(help.contains("/version"));
         assert!(help.contains("/export [file]"));
-        assert!(help.contains("/session [list|switch <session-id>|fork [branch-name]]"));
+        assert!(help.contains("/session [list|switch <session-id>]"));
         assert_eq!(slash_command_specs().len(), 15);
         assert_eq!(resume_supported_slash_commands().len(), 11);
     }
 
     #[test]
     fn compacts_sessions_via_slash_command() {
-        let mut session = Session::new();
-        session.messages = vec![
-            ConversationMessage::user_text("a ".repeat(200)),
-            ConversationMessage::assistant(vec![ContentBlock::Text {
-                text: "b ".repeat(200),
-            }]),
-            ConversationMessage::tool_result("1", "bash", "ok ".repeat(200), false),
-            ConversationMessage::assistant(vec![ContentBlock::Text {
-                text: "recent".to_string(),
-            }]),
-        ];
+        let session = Session {
+            version: 1,
+            messages: vec![
+                ConversationMessage::user_text("a ".repeat(200)),
+                ConversationMessage::assistant(vec![ContentBlock::Text {
+                    text: "b ".repeat(200),
+                }]),
+                ConversationMessage::tool_result("1", "bash", "ok ".repeat(200), false),
+                ConversationMessage::assistant(vec![ContentBlock::Text {
+                    text: "recent".to_string(),
+                }]),
+            ],
+        };
 
         let result = handle_slash_command(
             "/compact",
@@ -460,12 +455,6 @@ mod tests {
             CompactionConfig::default()
         )
         .is_none());
-        assert!(handle_slash_command(
-            "/resume session.jsonl",
-            &session,
-            CompactionConfig::default()
-        )
-        .is_none());
         assert!(handle_slash_command("/config", &session, CompactionConfig::default()).is_none());
         assert!(
             handle_slash_command("/config env", &session, CompactionConfig::default()).is_none()

rust/crates/runtime/src/compact.rs

@@ -99,14 +99,13 @@ pub fn compact_session(session: &Session, config: CompactionConfig) -> Compactio
     }];
     compacted_messages.extend(preserved);
 
-    let mut compacted_session = session.clone();
-    compacted_session.messages = compacted_messages;
-    compacted_session.record_compaction(summary.clone(), removed.len());
-
     CompactionResult {
         summary,
         formatted_summary,
-        compacted_session,
+        compacted_session: Session {
+            version: session.version,
+            messages: compacted_messages,
+        },
         removed_message_count: removed.len(),
     }
 }
@@ -391,8 +390,10 @@ mod tests {
 
     #[test]
     fn leaves_small_sessions_unchanged() {
-        let mut session = Session::new();
-        session.messages = vec![ConversationMessage::user_text("hello")];
+        let session = Session {
+            version: 1,
+            messages: vec![ConversationMessage::user_text("hello")],
+        };
 
         let result = compact_session(&session, CompactionConfig::default());
         assert_eq!(result.removed_message_count, 0);
@@ -403,21 +404,23 @@ mod tests {
 
     #[test]
     fn compacts_older_messages_into_a_system_summary() {
-        let mut session = Session::new();
-        session.messages = vec![
-            ConversationMessage::user_text("one ".repeat(200)),
-            ConversationMessage::assistant(vec![ContentBlock::Text {
-                text: "two ".repeat(200),
-            }]),
-            ConversationMessage::tool_result("1", "bash", "ok ".repeat(200), false),
-            ConversationMessage {
-                role: MessageRole::Assistant,
-                blocks: vec![ContentBlock::Text {
-                    text: "recent".to_string(),
-                }],
-                usage: None,
-            },
-        ];
+        let session = Session {
+            version: 1,
+            messages: vec![
+                ConversationMessage::user_text("one ".repeat(200)),
+                ConversationMessage::assistant(vec![ContentBlock::Text {
+                    text: "two ".repeat(200),
+                }]),
+                ConversationMessage::tool_result("1", "bash", "ok ".repeat(200), false),
+                ConversationMessage {
+                    role: MessageRole::Assistant,
+                    blocks: vec![ContentBlock::Text {
+                        text: "recent".to_string(),
+                    }],
+                    usage: None,
+                },
+            ],
+        };
 
         let result = compact_session(
             &session,

rust/crates/runtime/src/conversation.rs

@@ -118,7 +118,7 @@ where
             tool_executor,
             permission_policy,
             system_prompt,
-            RuntimeFeatureConfig::default(),
+            &RuntimeFeatureConfig::default(),
         )
     }
 
@@ -129,7 +129,7 @@ where
         tool_executor: T,
         permission_policy: PermissionPolicy,
         system_prompt: Vec<String>,
-        feature_config: RuntimeFeatureConfig,
+        feature_config: &RuntimeFeatureConfig,
     ) -> Self {
         let usage_tracker = UsageTracker::from_session(&session);
         Self {
@@ -140,7 +140,7 @@ where
             system_prompt,
             max_iterations: usize::MAX,
             usage_tracker,
-            hook_runner: HookRunner::from_feature_config(&feature_config),
+            hook_runner: HookRunner::from_feature_config(feature_config),
         }
     }
 
@@ -156,8 +156,8 @@ where
         mut prompter: Option<&mut dyn PermissionPrompter>,
     ) -> Result<TurnSummary, RuntimeError> {
         self.session
-            .push_user_text(user_input.into())
-            .map_err(|error| RuntimeError::new(error.to_string()))?;
+            .messages
+            .push(ConversationMessage::user_text(user_input.into()));
 
         let mut assistant_messages = Vec::new();
         let mut tool_results = Vec::new();
@@ -191,9 +191,7 @@ where
                 })
                 .collect::<Vec<_>>();
 
-            self.session
-                .push_message(assistant_message.clone())
-                .map_err(|error| RuntimeError::new(error.to_string()))?;
+            self.session.messages.push(assistant_message.clone());
             assistant_messages.push(assistant_message);
 
             if pending_tool_uses.is_empty() {
@@ -251,9 +249,7 @@ where
                         ConversationMessage::tool_result(tool_use_id, tool_name, reason, true)
                     }
                 };
-                self.session
-                    .push_message(result_message.clone())
-                    .map_err(|error| RuntimeError::new(error.to_string()))?;
+                self.session.messages.push(result_message.clone());
                 tool_results.push(result_message);
             }
         }
@@ -286,11 +282,6 @@ where
         &self.session
     }
 
-    #[must_use]
-    pub fn fork_session(&self, branch_name: Option<String>) -> Session {
-        self.session.fork(branch_name)
-    }
-
     #[must_use]
     pub fn into_session(self) -> Session {
         self.session
@@ -417,9 +408,7 @@ mod tests {
     use crate::prompt::{ProjectContext, SystemPromptBuilder};
     use crate::session::{ContentBlock, MessageRole, Session};
     use crate::usage::TokenUsage;
-    use std::fs;
     use std::path::PathBuf;
-    use std::time::{SystemTime, UNIX_EPOCH};
 
     struct ScriptedApiClient {
         call_count: usize,
@@ -620,7 +609,7 @@ mod tests {
             }),
             PermissionPolicy::new(PermissionMode::DangerFullAccess),
             vec!["system".to_string()],
-            RuntimeFeatureConfig::default().with_hooks(RuntimeHookConfig::new(
+            &RuntimeFeatureConfig::default().with_hooks(RuntimeHookConfig::new(
                 vec![shell_snippet("printf 'blocked by hook'; exit 2")],
                 Vec::new(),
             )),
@@ -686,7 +675,7 @@ mod tests {
             StaticToolExecutor::new().register("add", |_input| Ok("4".to_string())),
             PermissionPolicy::new(PermissionMode::DangerFullAccess),
             vec!["system".to_string()],
-            RuntimeFeatureConfig::default().with_hooks(RuntimeHookConfig::new(
+            &RuntimeFeatureConfig::default().with_hooks(RuntimeHookConfig::new(
                 vec![shell_snippet("printf 'pre hook ran'")],
                 vec![shell_snippet("printf 'post hook ran'")],
             )),
@@ -708,7 +697,7 @@ mod tests {
             "post hook should preserve non-error result: {output:?}"
         );
         assert!(
-            output.contains("4"),
+            output.contains('4'),
             "tool output missing value: {output:?}"
         );
         assert!(
@@ -798,86 +787,6 @@ mod tests {
             result.compacted_session.messages[0].role,
             MessageRole::System
         );
-        assert_eq!(
-            result.compacted_session.session_id,
-            runtime.session().session_id
-        );
-        assert!(result.compacted_session.compaction.is_some());
-    }
-
-    #[test]
-    fn persists_conversation_turn_messages_to_jsonl_session() {
-        struct SimpleApi;
-        impl ApiClient for SimpleApi {
-            fn stream(
-                &mut self,
-                _request: ApiRequest,
-            ) -> Result<Vec<AssistantEvent>, RuntimeError> {
-                Ok(vec![
-                    AssistantEvent::TextDelta("done".to_string()),
-                    AssistantEvent::MessageStop,
-                ])
-            }
-        }
-
-        let path = temp_session_path("persisted-turn");
-        let session = Session::new().with_persistence_path(path.clone());
-        let mut runtime = ConversationRuntime::new(
-            session,
-            SimpleApi,
-            StaticToolExecutor::new(),
-            PermissionPolicy::new(PermissionMode::DangerFullAccess),
-            vec!["system".to_string()],
-        );
-
-        runtime
-            .run_turn("persist this turn", None)
-            .expect("turn should succeed");
-
-        let restored = Session::load_from_path(&path).expect("persisted session should reload");
-        fs::remove_file(&path).expect("temp session file should be removable");
-
-        assert_eq!(restored.messages.len(), 2);
-        assert_eq!(restored.messages[0].role, MessageRole::User);
-        assert_eq!(restored.messages[1].role, MessageRole::Assistant);
-        assert_eq!(restored.session_id, runtime.session().session_id);
-    }
-
-    #[test]
-    fn forks_runtime_session_without_mutating_original() {
-        let mut session = Session::new();
-        session
-            .push_user_text("branch me")
-            .expect("message should append");
-
-        let runtime = ConversationRuntime::new(
-            session.clone(),
-            ScriptedApiClient { call_count: 0 },
-            StaticToolExecutor::new(),
-            PermissionPolicy::new(PermissionMode::DangerFullAccess),
-            vec!["system".to_string()],
-        );
-
-        let forked = runtime.fork_session(Some("alt-path".to_string()));
-
-        assert_eq!(forked.messages, session.messages);
-        assert_ne!(forked.session_id, session.session_id);
-        assert_eq!(
-            forked
-                .fork
-                .as_ref()
-                .map(|fork| (fork.parent_session_id.as_str(), fork.branch_name.as_deref())),
-            Some((session.session_id.as_str(), Some("alt-path")))
-        );
-        assert!(runtime.session().fork.is_none());
-    }
-
-    fn temp_session_path(label: &str) -> PathBuf {
-        let nanos = SystemTime::now()
-            .duration_since(UNIX_EPOCH)
-            .expect("system time should be after epoch")
-            .as_nanos();
-        std::env::temp_dir().join(format!("runtime-conversation-{label}-{nanos}.json"))
     }
 
     #[cfg(windows)]

rust/crates/runtime/src/hooks.rs

@@ -51,6 +51,16 @@ pub struct HookRunner {
     config: RuntimeHookConfig,
 }
 
+#[derive(Debug, Clone, Copy)]
+struct HookCommandRequest<'a> {
+    event: HookEvent,
+    tool_name: &'a str,
+    tool_input: &'a str,
+    tool_output: Option<&'a str>,
+    is_error: bool,
+    payload: &'a str,
+}
+
 impl HookRunner {
     #[must_use]
     pub fn new(config: RuntimeHookConfig) -> Self {
@@ -118,14 +128,16 @@ impl HookRunner {
         let mut messages = Vec::new();
 
         for command in commands {
-            match self.run_command(
+            match Self::run_command(
                 command,
-                event,
-                tool_name,
-                tool_input,
-                tool_output,
-                is_error,
-                &payload,
+                HookCommandRequest {
+                    event,
+                    tool_name,
+                    tool_input,
+                    tool_output,
+                    is_error,
+                    payload: &payload,
+                },
             ) {
                 HookCommandOutcome::Allow { message } => {
                     if let Some(message) = message {
@@ -149,29 +161,23 @@ impl HookRunner {
         HookRunResult::allow(messages)
     }
 
-    fn run_command(
-        &self,
-        command: &str,
-        event: HookEvent,
-        tool_name: &str,
-        tool_input: &str,
-        tool_output: Option<&str>,
-        is_error: bool,
-        payload: &str,
-    ) -> HookCommandOutcome {
+    fn run_command(command: &str, request: HookCommandRequest<'_>) -> HookCommandOutcome {
         let mut child = shell_command(command);
         child.stdin(std::process::Stdio::piped());
         child.stdout(std::process::Stdio::piped());
         child.stderr(std::process::Stdio::piped());
-        child.env("HOOK_EVENT", event.as_str());
-        child.env("HOOK_TOOL_NAME", tool_name);
-        child.env("HOOK_TOOL_INPUT", tool_input);
-        child.env("HOOK_TOOL_IS_ERROR", if is_error { "1" } else { "0" });
-        if let Some(tool_output) = tool_output {
+        child.env("HOOK_EVENT", request.event.as_str());
+        child.env("HOOK_TOOL_NAME", request.tool_name);
+        child.env("HOOK_TOOL_INPUT", request.tool_input);
+        child.env(
+            "HOOK_TOOL_IS_ERROR",
+            if request.is_error { "1" } else { "0" },
+        );
+        if let Some(tool_output) = request.tool_output {
             child.env("HOOK_TOOL_OUTPUT", tool_output);
         }
 
-        match child.output_with_stdin(payload.as_bytes()) {
+        match child.output_with_stdin(request.payload.as_bytes()) {
             Ok(output) => {
                 let stdout = String::from_utf8_lossy(&output.stdout).trim().to_string();
                 let stderr = String::from_utf8_lossy(&output.stderr).trim().to_string();
@@ -189,16 +195,18 @@ impl HookRunner {
                     },
                     None => HookCommandOutcome::Warn {
                         message: format!(
-                            "{} hook `{command}` terminated by signal while handling `{tool_name}`",
-                            event.as_str()
+                            "{} hook `{command}` terminated by signal while handling `{}`",
+                            request.event.as_str(),
+                            request.tool_name
                         ),
                     },
                 }
             }
             Err(error) => HookCommandOutcome::Warn {
                 message: format!(
-                    "{} hook `{command}` failed to start for `{tool_name}`: {error}",
-                    event.as_str()
+                    "{} hook `{command}` failed to start for `{}`: {error}",
+                    request.event.as_str(),
+                    request.tool_name
                 ),
             },
         }

rust/crates/runtime/src/lib.rs

@@ -76,10 +76,7 @@ pub use remote::{
     RemoteSessionContext, UpstreamProxyBootstrap, UpstreamProxyState, DEFAULT_REMOTE_BASE_URL,
     DEFAULT_SESSION_TOKEN_PATH, DEFAULT_SYSTEM_CA_BUNDLE, NO_PROXY_HOSTS, UPSTREAM_PROXY_ENV_KEYS,
 };
-pub use session::{
-    ContentBlock, ConversationMessage, MessageRole, Session, SessionCompaction, SessionError,
-    SessionFork,
-};
+pub use session::{ContentBlock, ConversationMessage, MessageRole, Session, SessionError};
 pub use usage::{
     format_usd, pricing_for_model, ModelPricing, TokenUsage, UsageCostEstimate, UsageTracker,
 };

rust/crates/runtime/src/session.rs

@@ -1,19 +1,11 @@
 use std::collections::BTreeMap;
 use std::fmt::{Display, Formatter};
-use std::fs::{self, OpenOptions};
-use std::io::Write;
-use std::path::{Path, PathBuf};
-use std::sync::atomic::{AtomicU64, Ordering};
-use std::time::{SystemTime, UNIX_EPOCH};
+use std::fs;
+use std::path::Path;
 
 use crate::json::{JsonError, JsonValue};
 use crate::usage::TokenUsage;
 
-const SESSION_VERSION: u32 = 1;
-const ROTATE_AFTER_BYTES: u64 = 256 * 1024;
-const MAX_ROTATED_FILES: usize = 3;
-static SESSION_ID_COUNTER: AtomicU64 = AtomicU64::new(0);
-
 #[derive(Debug, Clone, Copy, PartialEq, Eq)]
 pub enum MessageRole {
     System,
@@ -48,49 +40,11 @@ pub struct ConversationMessage {
 }
 
 #[derive(Debug, Clone, PartialEq, Eq)]
-pub struct SessionCompaction {
-    pub count: u32,
-    pub removed_message_count: usize,
-    pub summary: String,
-}
-
-#[derive(Debug, Clone, PartialEq, Eq)]
-pub struct SessionFork {
-    pub parent_session_id: String,
-    pub branch_name: Option<String>,
-}
-
-#[derive(Debug, Clone, PartialEq, Eq)]
-struct SessionPersistence {
-    path: PathBuf,
-}
-
-#[derive(Debug, Clone)]
 pub struct Session {
     pub version: u32,
-    pub session_id: String,
-    pub created_at_ms: u64,
-    pub updated_at_ms: u64,
     pub messages: Vec<ConversationMessage>,
-    pub compaction: Option<SessionCompaction>,
-    pub fork: Option<SessionFork>,
-    persistence: Option<SessionPersistence>,
 }
 
-impl PartialEq for Session {
-    fn eq(&self, other: &Self) -> bool {
-        self.version == other.version
-            && self.session_id == other.session_id
-            && self.created_at_ms == other.created_at_ms
-            && self.updated_at_ms == other.updated_at_ms
-            && self.messages == other.messages
-            && self.compaction == other.compaction
-            && self.fork == other.fork
-    }
-}
-
-impl Eq for Session {}
-
 #[derive(Debug)]
 pub enum SessionError {
     Io(std::io::Error),
@@ -125,84 +79,20 @@ impl From<JsonError> for SessionError {
 impl Session {
     #[must_use]
     pub fn new() -> Self {
-        let now = current_time_millis();
         Self {
-            version: SESSION_VERSION,
-            session_id: generate_session_id(),
-            created_at_ms: now,
-            updated_at_ms: now,
+            version: 1,
             messages: Vec::new(),
-            compaction: None,
-            fork: None,
-            persistence: None,
         }
     }
 
-    #[must_use]
-    pub fn with_persistence_path(mut self, path: impl Into<PathBuf>) -> Self {
-        self.persistence = Some(SessionPersistence { path: path.into() });
-        self
-    }
-
-    #[must_use]
-    pub fn persistence_path(&self) -> Option<&Path> {
-        self.persistence.as_ref().map(|value| value.path.as_path())
-    }
-
     pub fn save_to_path(&self, path: impl AsRef<Path>) -> Result<(), SessionError> {
-        let path = path.as_ref();
-        rotate_session_file_if_needed(path)?;
-        write_atomic(path, &self.render_jsonl_snapshot())?;
-        cleanup_rotated_logs(path)?;
+        fs::write(path, self.to_json().render())?;
         Ok(())
     }
 
     pub fn load_from_path(path: impl AsRef<Path>) -> Result<Self, SessionError> {
-        let path = path.as_ref();
         let contents = fs::read_to_string(path)?;
-        let session = match JsonValue::parse(&contents) {
-            Ok(value) => Self::from_json(&value)?,
-            Err(_) => Self::from_jsonl(&contents)?,
-        };
-        Ok(session.with_persistence_path(path.to_path_buf()))
-    }
-
-    pub fn push_message(&mut self, message: ConversationMessage) -> Result<(), SessionError> {
-        self.touch();
-        self.messages.push(message.clone());
-        self.append_persisted_message(&message)
-    }
-
-    pub fn push_user_text(&mut self, text: impl Into<String>) -> Result<(), SessionError> {
-        self.push_message(ConversationMessage::user_text(text))
-    }
-
-    pub fn record_compaction(&mut self, summary: impl Into<String>, removed_message_count: usize) {
-        self.touch();
-        let count = self.compaction.as_ref().map_or(1, |value| value.count + 1);
-        self.compaction = Some(SessionCompaction {
-            count,
-            removed_message_count,
-            summary: summary.into(),
-        });
-    }
-
-    #[must_use]
-    pub fn fork(&self, branch_name: Option<String>) -> Self {
-        let now = current_time_millis();
-        Self {
-            version: self.version,
-            session_id: generate_session_id(),
-            created_at_ms: now,
-            updated_at_ms: now,
-            messages: self.messages.clone(),
-            compaction: self.compaction.clone(),
-            fork: Some(SessionFork {
-                parent_session_id: self.session_id.clone(),
-                branch_name: normalize_optional_string(branch_name),
-            }),
-            persistence: None,
-        }
+        Self::from_json(&JsonValue::parse(&contents)?)
     }
 
     #[must_use]
@@ -212,18 +102,6 @@ impl Session {
             "version".to_string(),
             JsonValue::Number(i64::from(self.version)),
         );
-        object.insert(
-            "session_id".to_string(),
-            JsonValue::String(self.session_id.clone()),
-        );
-        object.insert(
-            "created_at_ms".to_string(),
-            JsonValue::Number(i64_from_u64(self.created_at_ms, "created_at_ms")),
-        );
-        object.insert(
-            "updated_at_ms".to_string(),
-            JsonValue::Number(i64_from_u64(self.updated_at_ms, "updated_at_ms")),
-        );
         object.insert(
             "messages".to_string(),
             JsonValue::Array(
@@ -233,12 +111,6 @@ impl Session {
                     .collect(),
             ),
         );
-        if let Some(compaction) = &self.compaction {
-            object.insert("compaction".to_string(), compaction.to_json());
-        }
-        if let Some(fork) = &self.fork {
-            object.insert("fork".to_string(), fork.to_json());
-        }
         JsonValue::Object(object)
     }
 
@@ -259,179 +131,7 @@ impl Session {
             .iter()
             .map(ConversationMessage::from_json)
             .collect::<Result<Vec<_>, _>>()?;
-        let now = current_time_millis();
-        let session_id = object
-            .get("session_id")
-            .and_then(JsonValue::as_str)
-            .map(ToOwned::to_owned)
-            .unwrap_or_else(generate_session_id);
-        let created_at_ms = object
-            .get("created_at_ms")
-            .map(|value| required_u64_from_value(value, "created_at_ms"))
-            .transpose()?
-            .unwrap_or(now);
-        let updated_at_ms = object
-            .get("updated_at_ms")
-            .map(|value| required_u64_from_value(value, "updated_at_ms"))
-            .transpose()?
-            .unwrap_or(created_at_ms);
-        let compaction = object
-            .get("compaction")
-            .map(SessionCompaction::from_json)
-            .transpose()?;
-        let fork = object.get("fork").map(SessionFork::from_json).transpose()?;
-        Ok(Self {
-            version,
-            session_id,
-            created_at_ms,
-            updated_at_ms,
-            messages,
-            compaction,
-            fork,
-            persistence: None,
-        })
-    }
-
-    fn from_jsonl(contents: &str) -> Result<Self, SessionError> {
-        let mut version = SESSION_VERSION;
-        let mut session_id = None;
-        let mut created_at_ms = None;
-        let mut updated_at_ms = None;
-        let mut messages = Vec::new();
-        let mut compaction = None;
-        let mut fork = None;
-
-        for (line_number, raw_line) in contents.lines().enumerate() {
-            let line = raw_line.trim();
-            if line.is_empty() {
-                continue;
-            }
-            let value = JsonValue::parse(line).map_err(|error| {
-                SessionError::Format(format!(
-                    "invalid JSONL record at line {}: {}",
-                    line_number + 1,
-                    error
-                ))
-            })?;
-            let object = value.as_object().ok_or_else(|| {
-                SessionError::Format(format!(
-                    "JSONL record at line {} must be an object",
-                    line_number + 1
-                ))
-            })?;
-            match object
-                .get("type")
-                .and_then(JsonValue::as_str)
-                .ok_or_else(|| {
-                    SessionError::Format(format!(
-                        "JSONL record at line {} missing type",
-                        line_number + 1
-                    ))
-                })? {
-                "session_meta" => {
-                    version = required_u32(object, "version")?;
-                    session_id = Some(required_string(object, "session_id")?);
-                    created_at_ms = Some(required_u64(object, "created_at_ms")?);
-                    updated_at_ms = Some(required_u64(object, "updated_at_ms")?);
-                    fork = object.get("fork").map(SessionFork::from_json).transpose()?;
-                }
-                "message" => {
-                    let message_value = object.get("message").ok_or_else(|| {
-                        SessionError::Format(format!(
-                            "JSONL record at line {} missing message",
-                            line_number + 1
-                        ))
-                    })?;
-                    messages.push(ConversationMessage::from_json(message_value)?);
-                }
-                "compaction" => {
-                    compaction = Some(SessionCompaction::from_json(&JsonValue::Object(
-                        object.clone(),
-                    ))?);
-                }
-                other => {
-                    return Err(SessionError::Format(format!(
-                        "unsupported JSONL record type at line {}: {other}",
-                        line_number + 1
-                    )))
-                }
-            }
-        }
-
-        let now = current_time_millis();
-        Ok(Self {
-            version,
-            session_id: session_id.unwrap_or_else(generate_session_id),
-            created_at_ms: created_at_ms.unwrap_or(now),
-            updated_at_ms: updated_at_ms.unwrap_or(created_at_ms.unwrap_or(now)),
-            messages,
-            compaction,
-            fork,
-            persistence: None,
-        })
-    }
-
-    fn render_jsonl_snapshot(&self) -> String {
-        let mut lines = vec![self.meta_record().render()];
-        if let Some(compaction) = &self.compaction {
-            lines.push(compaction.to_jsonl_record().render());
-        }
-        lines.extend(
-            self.messages
-                .iter()
-                .map(|message| message_record(message).render()),
-        );
-        let mut rendered = lines.join("\n");
-        rendered.push('\n');
-        rendered
-    }
-
-    fn append_persisted_message(&self, message: &ConversationMessage) -> Result<(), SessionError> {
-        let Some(path) = self.persistence_path() else {
-            return Ok(());
-        };
-
-        let needs_bootstrap = !path.exists() || fs::metadata(path)?.len() == 0;
-        if needs_bootstrap {
-            self.save_to_path(path)?;
-            return Ok(());
-        }
-
-        let mut file = OpenOptions::new().append(true).open(path)?;
-        writeln!(file, "{}", message_record(message).render())?;
-        Ok(())
-    }
-
-    fn meta_record(&self) -> JsonValue {
-        let mut object = BTreeMap::new();
-        object.insert(
-            "type".to_string(),
-            JsonValue::String("session_meta".to_string()),
-        );
-        object.insert(
-            "version".to_string(),
-            JsonValue::Number(i64::from(self.version)),
-        );
-        object.insert(
-            "session_id".to_string(),
-            JsonValue::String(self.session_id.clone()),
-        );
-        object.insert(
-            "created_at_ms".to_string(),
-            JsonValue::Number(i64_from_u64(self.created_at_ms, "created_at_ms")),
-        );
-        object.insert(
-            "updated_at_ms".to_string(),
-            JsonValue::Number(i64_from_u64(self.updated_at_ms, "updated_at_ms")),
-        );
-        if let Some(fork) = &self.fork {
-            object.insert("fork".to_string(), fork.to_json());
-        }
-        JsonValue::Object(object)
-    }
-
-    fn touch(&mut self) {
-        self.updated_at_ms = current_time_millis();
+        Ok(Self { version, messages })
     }
 }
 
@@ -624,92 +324,6 @@ impl ContentBlock {
     }
 }
 
-impl SessionCompaction {
-    #[must_use]
-    pub fn to_json(&self) -> JsonValue {
-        let mut object = BTreeMap::new();
-        object.insert(
-            "count".to_string(),
-            JsonValue::Number(i64::from(self.count)),
-        );
-        object.insert(
-            "removed_message_count".to_string(),
-            JsonValue::Number(i64_from_usize(
-                self.removed_message_count,
-                "removed_message_count",
-            )),
-        );
-        object.insert(
-            "summary".to_string(),
-            JsonValue::String(self.summary.clone()),
-        );
-        JsonValue::Object(object)
-    }
-
-    #[must_use]
-    pub fn to_jsonl_record(&self) -> JsonValue {
-        let mut object = self
-            .to_json()
-            .as_object()
-            .cloned()
-            .expect("compaction should render to object");
-        object.insert(
-            "type".to_string(),
-            JsonValue::String("compaction".to_string()),
-        );
-        JsonValue::Object(object)
-    }
-
-    fn from_json(value: &JsonValue) -> Result<Self, SessionError> {
-        let object = value
-            .as_object()
-            .ok_or_else(|| SessionError::Format("compaction must be an object".to_string()))?;
-        Ok(Self {
-            count: required_u32(object, "count")?,
-            removed_message_count: required_usize(object, "removed_message_count")?,
-            summary: required_string(object, "summary")?,
-        })
-    }
-}
-
-impl SessionFork {
-    #[must_use]
-    pub fn to_json(&self) -> JsonValue {
-        let mut object = BTreeMap::new();
-        object.insert(
-            "parent_session_id".to_string(),
-            JsonValue::String(self.parent_session_id.clone()),
-        );
-        if let Some(branch_name) = &self.branch_name {
-            object.insert(
-                "branch_name".to_string(),
-                JsonValue::String(branch_name.clone()),
-            );
-        }
-        JsonValue::Object(object)
-    }
-
-    fn from_json(value: &JsonValue) -> Result<Self, SessionError> {
-        let object = value
-            .as_object()
-            .ok_or_else(|| SessionError::Format("fork metadata must be an object".to_string()))?;
-        Ok(Self {
-            parent_session_id: required_string(object, "parent_session_id")?,
-            branch_name: object
-                .get("branch_name")
-                .and_then(JsonValue::as_str)
-                .map(ToOwned::to_owned),
-        })
-    }
-}
-
-fn message_record(message: &ConversationMessage) -> JsonValue {
-    let mut object = BTreeMap::new();
-    object.insert("type".to_string(), JsonValue::String("message".to_string()));
-    object.insert("message".to_string(), message.to_json());
-    JsonValue::Object(object)
-}
-
 fn usage_to_json(usage: TokenUsage) -> JsonValue {
     let mut object = BTreeMap::new();
     object.insert(
@@ -762,155 +376,22 @@ fn required_u32(object: &BTreeMap<String, JsonValue>, key: &str) -> Result<u32,
     u32::try_from(value).map_err(|_| SessionError::Format(format!("{key} out of range")))
 }
 
-fn required_u64(object: &BTreeMap<String, JsonValue>, key: &str) -> Result<u64, SessionError> {
-    let value = object
-        .get(key)
-        .ok_or_else(|| SessionError::Format(format!("missing {key}")))?;
-    required_u64_from_value(value, key)
-}
-
-fn required_u64_from_value(value: &JsonValue, key: &str) -> Result<u64, SessionError> {
-    let value = value
-        .as_i64()
-        .ok_or_else(|| SessionError::Format(format!("missing {key}")))?;
-    u64::try_from(value).map_err(|_| SessionError::Format(format!("{key} out of range")))
-}
-
-fn required_usize(object: &BTreeMap<String, JsonValue>, key: &str) -> Result<usize, SessionError> {
-    let value = object
-        .get(key)
-        .and_then(JsonValue::as_i64)
-        .ok_or_else(|| SessionError::Format(format!("missing {key}")))?;
-    usize::try_from(value).map_err(|_| SessionError::Format(format!("{key} out of range")))
-}
-
-fn i64_from_u64(value: u64, key: &str) -> i64 {
-    i64::try_from(value).unwrap_or_else(|_| panic!("{key} out of range for JSON number"))
-}
-
-fn i64_from_usize(value: usize, key: &str) -> i64 {
-    i64::try_from(value).unwrap_or_else(|_| panic!("{key} out of range for JSON number"))
-}
-
-fn normalize_optional_string(value: Option<String>) -> Option<String> {
-    value.and_then(|value| {
-        let trimmed = value.trim();
-        if trimmed.is_empty() {
-            None
-        } else {
-            Some(trimmed.to_string())
-        }
-    })
-}
-
-fn current_time_millis() -> u64 {
-    SystemTime::now()
-        .duration_since(UNIX_EPOCH)
-        .map(|duration| duration.as_millis() as u64)
-        .unwrap_or_default()
-}
-
-fn generate_session_id() -> String {
-    let millis = current_time_millis();
-    let counter = SESSION_ID_COUNTER.fetch_add(1, Ordering::Relaxed);
-    format!("session-{millis}-{counter}")
-}
-
-fn write_atomic(path: &Path, contents: &str) -> Result<(), SessionError> {
-    if let Some(parent) = path.parent() {
-        fs::create_dir_all(parent)?;
-    }
-    let temp_path = temporary_path_for(path);
-    fs::write(&temp_path, contents)?;
-    fs::rename(temp_path, path)?;
-    Ok(())
-}
-
-fn temporary_path_for(path: &Path) -> PathBuf {
-    let file_name = path
-        .file_name()
-        .and_then(|value| value.to_str())
-        .unwrap_or("session");
-    path.with_file_name(format!(
-        "{file_name}.tmp-{}-{}",
-        current_time_millis(),
-        SESSION_ID_COUNTER.fetch_add(1, Ordering::Relaxed)
-    ))
-}
-
-fn rotate_session_file_if_needed(path: &Path) -> Result<(), SessionError> {
-    let Ok(metadata) = fs::metadata(path) else {
-        return Ok(());
-    };
-    if metadata.len() < ROTATE_AFTER_BYTES {
-        return Ok(());
-    }
-    let rotated_path = rotated_log_path(path);
-    fs::rename(path, rotated_path)?;
-    Ok(())
-}
-
-fn rotated_log_path(path: &Path) -> PathBuf {
-    let stem = path
-        .file_stem()
-        .and_then(|value| value.to_str())
-        .unwrap_or("session");
-    path.with_file_name(format!("{stem}.rot-{}.jsonl", current_time_millis()))
-}
-
-fn cleanup_rotated_logs(path: &Path) -> Result<(), SessionError> {
-    let Some(parent) = path.parent() else {
-        return Ok(());
-    };
-    let stem = path
-        .file_stem()
-        .and_then(|value| value.to_str())
-        .unwrap_or("session");
-    let prefix = format!("{stem}.rot-");
-    let mut rotated_paths = fs::read_dir(parent)?
-        .filter_map(Result::ok)
-        .map(|entry| entry.path())
-        .filter(|entry_path| {
-            entry_path
-                .file_name()
-                .and_then(|value| value.to_str())
-                .is_some_and(|name| name.starts_with(&prefix) && name.ends_with(".jsonl"))
-        })
-        .collect::<Vec<_>>();
-
-    rotated_paths.sort_by_key(|entry_path| {
-        fs::metadata(entry_path)
-            .and_then(|metadata| metadata.modified())
-            .unwrap_or(UNIX_EPOCH)
-    });
-
-    let remove_count = rotated_paths.len().saturating_sub(MAX_ROTATED_FILES);
-    for stale_path in rotated_paths.into_iter().take(remove_count) {
-        fs::remove_file(stale_path)?;
-    }
-    Ok(())
-}
-
 #[cfg(test)]
 mod tests {
-    use super::{
-        cleanup_rotated_logs, rotate_session_file_if_needed, ContentBlock, ConversationMessage,
-        MessageRole, Session, SessionFork,
-    };
-    use crate::json::JsonValue;
+    use super::{ContentBlock, ConversationMessage, MessageRole, Session};
     use crate::usage::TokenUsage;
     use std::fs;
-    use std::path::PathBuf;
     use std::time::{SystemTime, UNIX_EPOCH};
 
     #[test]
-    fn persists_and_restores_session_jsonl() {
+    fn persists_and_restores_session_json() {
         let mut session = Session::new();
         session
-            .push_user_text("hello")
-            .expect("user message should append");
+            .messages
+            .push(ConversationMessage::user_text("hello"));
         session
-            .push_message(ConversationMessage::assistant_with_usage(
+            .messages
+            .push(ConversationMessage::assistant_with_usage(
                 vec![
                     ContentBlock::Text {
                         text: "thinking".to_string(),
@@ -927,15 +408,16 @@ mod tests {
                     cache_creation_input_tokens: 1,
                     cache_read_input_tokens: 2,
                 }),
-            ))
-            .expect("assistant message should append");
-        session
-            .push_message(ConversationMessage::tool_result(
-                "tool-1", "bash", "hi", false,
-            ))
-            .expect("tool result should append");
+            ));
+        session.messages.push(ConversationMessage::tool_result(
+            "tool-1", "bash", "hi", false,
+        ));
 
-        let path = temp_session_path("jsonl");
+        let nanos = SystemTime::now()
+            .duration_since(UNIX_EPOCH)
+            .expect("system time should be after epoch")
+            .as_nanos();
+        let path = std::env::temp_dir().join(format!("runtime-session-{nanos}.json"));
         session.save_to_path(&path).expect("session should save");
         let restored = Session::load_from_path(&path).expect("session should load");
         fs::remove_file(&path).expect("temp file should be removable");
@@ -946,157 +428,5 @@ mod tests {
             restored.messages[1].usage.expect("usage").total_tokens(),
             17
         );
-        assert_eq!(restored.session_id, session.session_id);
-    }
-
-    #[test]
-    fn loads_legacy_session_json_object() {
-        let path = temp_session_path("legacy");
-        let legacy = JsonValue::Object(
-            [
-                ("version".to_string(), JsonValue::Number(1)),
-                (
-                    "messages".to_string(),
-                    JsonValue::Array(vec![ConversationMessage::user_text("legacy").to_json()]),
-                ),
-            ]
-            .into_iter()
-            .collect(),
-        );
-        fs::write(&path, legacy.render()).expect("legacy file should write");
-
-        let restored = Session::load_from_path(&path).expect("legacy session should load");
-        fs::remove_file(&path).expect("temp file should be removable");
-
-        assert_eq!(restored.messages.len(), 1);
-        assert_eq!(
-            restored.messages[0],
-            ConversationMessage::user_text("legacy")
-        );
-        assert!(!restored.session_id.is_empty());
-    }
-
-    #[test]
-    fn appends_messages_to_persisted_jsonl_session() {
-        let path = temp_session_path("append");
-        let mut session = Session::new().with_persistence_path(path.clone());
-        session
-            .save_to_path(&path)
-            .expect("initial save should succeed");
-        session
-            .push_user_text("hi")
-            .expect("user append should succeed");
-        session
-            .push_message(ConversationMessage::assistant(vec![ContentBlock::Text {
-                text: "hello".to_string(),
-            }]))
-            .expect("assistant append should succeed");
-
-        let restored = Session::load_from_path(&path).expect("session should replay from jsonl");
-        fs::remove_file(&path).expect("temp file should be removable");
-
-        assert_eq!(restored.messages.len(), 2);
-        assert_eq!(restored.messages[0], ConversationMessage::user_text("hi"));
-    }
-
-    #[test]
-    fn persists_compaction_metadata() {
-        let path = temp_session_path("compaction");
-        let mut session = Session::new();
-        session
-            .push_user_text("before")
-            .expect("message should append");
-        session.record_compaction("summarized earlier work", 4);
-        session.save_to_path(&path).expect("session should save");
-
-        let restored = Session::load_from_path(&path).expect("session should load");
-        fs::remove_file(&path).expect("temp file should be removable");
-
-        let compaction = restored.compaction.expect("compaction metadata");
-        assert_eq!(compaction.count, 1);
-        assert_eq!(compaction.removed_message_count, 4);
-        assert!(compaction.summary.contains("summarized"));
-    }
-
-    #[test]
-    fn forks_sessions_with_branch_metadata_and_persists_it() {
-        let path = temp_session_path("fork");
-        let mut session = Session::new();
-        session
-            .push_user_text("before fork")
-            .expect("message should append");
-
-        let forked = session
-            .fork(Some("investigation".to_string()))
-            .with_persistence_path(path.clone());
-        forked
-            .save_to_path(&path)
-            .expect("forked session should save");
-
-        let restored = Session::load_from_path(&path).expect("forked session should load");
-        fs::remove_file(&path).expect("temp file should be removable");
-
-        assert_ne!(restored.session_id, session.session_id);
-        assert_eq!(
-            restored.fork,
-            Some(SessionFork {
-                parent_session_id: session.session_id,
-                branch_name: Some("investigation".to_string()),
-            })
-        );
-        assert_eq!(restored.messages, forked.messages);
-    }
-
-    #[test]
-    fn rotates_and_cleans_up_large_session_logs() {
-        let path = temp_session_path("rotation");
-        fs::write(&path, "x".repeat((super::ROTATE_AFTER_BYTES + 10) as usize))
-            .expect("oversized file should write");
-        rotate_session_file_if_needed(&path).expect("rotation should succeed");
-        assert!(
-            !path.exists(),
-            "original path should be rotated away before rewrite"
-        );
-
-        for _ in 0..5 {
-            let rotated = super::rotated_log_path(&path);
-            fs::write(&rotated, "old").expect("rotated file should write");
-        }
-        cleanup_rotated_logs(&path).expect("cleanup should succeed");
-
-        let rotated_count = rotation_files(&path).len();
-        assert!(rotated_count <= super::MAX_ROTATED_FILES);
-        for rotated in rotation_files(&path) {
-            fs::remove_file(rotated).expect("rotated file should be removable");
-        }
-    }
-
-    fn temp_session_path(label: &str) -> PathBuf {
-        let nanos = SystemTime::now()
-            .duration_since(UNIX_EPOCH)
-            .expect("system time should be after epoch")
-            .as_nanos();
-        std::env::temp_dir().join(format!("runtime-session-{label}-{nanos}.json"))
-    }
-
-    fn rotation_files(path: &PathBuf) -> Vec<PathBuf> {
-        let stem = path
-            .file_stem()
-            .and_then(|value| value.to_str())
-            .expect("temp path should have file stem")
-            .to_string();
-        fs::read_dir(path.parent().expect("temp path should have parent"))
-            .expect("temp dir should read")
-            .filter_map(Result::ok)
-            .map(|entry| entry.path())
-            .filter(|entry_path| {
-                entry_path
-                    .file_name()
-                    .and_then(|value| value.to_str())
-                    .is_some_and(|name| {
-                        name.starts_with(&format!("{stem}.rot-")) && name.ends_with(".jsonl")
-                    })
-            })
-            .collect()
     }
 }

rust/crates/runtime/src/usage.rs

@@ -286,19 +286,21 @@ mod tests {
 
     #[test]
     fn reconstructs_usage_from_session_messages() {
-        let mut session = Session::new();
-        session.messages = vec![ConversationMessage {
-            role: MessageRole::Assistant,
-            blocks: vec![ContentBlock::Text {
-                text: "done".to_string(),
+        let session = Session {
+            version: 1,
+            messages: vec![ConversationMessage {
+                role: MessageRole::Assistant,
+                blocks: vec![ContentBlock::Text {
+                    text: "done".to_string(),
+                }],
+                usage: Some(TokenUsage {
+                    input_tokens: 5,
+                    output_tokens: 2,
+                    cache_creation_input_tokens: 1,
+                    cache_read_input_tokens: 0,
+                }),
             }],
-            usage: Some(TokenUsage {
-                input_tokens: 5,
-                output_tokens: 2,
-                cache_creation_input_tokens: 1,
-                cache_read_input_tokens: 0,
-            }),
-        }];
+        };
 
         let tracker = UsageTracker::from_session(&session);
         assert_eq!(tracker.turns(), 1);

rust/crates/rusty-claude-cli/src/main.rs

@@ -9,11 +9,12 @@ use std::io::{self, Read, Write};
 use std::net::TcpListener;
 use std::path::{Path, PathBuf};
 use std::process::Command;
-use std::time::UNIX_EPOCH;
+use std::time::{SystemTime, UNIX_EPOCH};
 
 use api::{
-    resolve_startup_auth_source, AnthropicClient, AuthSource, ContentBlockDelta, InputContentBlock,
-    InputMessage, MessageRequest, MessageResponse, OutputContentBlock,
+    detect_provider_kind, max_tokens_for_model, resolve_model_alias, resolve_startup_auth_source,
+    AnthropicClient, AuthSource, ContentBlockDelta, InputContentBlock, InputMessage,
+    MessageRequest, MessageResponse, OutputContentBlock, ProviderClient, ProviderKind,
     StreamEvent as ApiStreamEvent, ToolChoice, ToolDefinition, ToolResultContentBlock,
 };
 
@@ -35,20 +36,11 @@ use serde_json::json;
 use tools::{execute_tool, mvp_tool_specs, ToolSpec};
 
 const DEFAULT_MODEL: &str = "claude-opus-4-6";
-fn max_tokens_for_model(model: &str) -> u32 {
-    if model.contains("opus") {
-        32_000
-    } else {
-        64_000
-    }
-}
 const DEFAULT_DATE: &str = "2026-03-31";
 const DEFAULT_OAUTH_CALLBACK_PORT: u16 = 4545;
 const VERSION: &str = env!("CARGO_PKG_VERSION");
 const BUILD_TARGET: Option<&str> = option_env!("TARGET");
 const GIT_SHA: Option<&str> = option_env!("GIT_SHA");
-const PRIMARY_SESSION_EXTENSION: &str = "jsonl";
-const LEGACY_SESSION_EXTENSION: &str = "json";
 
 type AllowedToolSet = BTreeSet<String>;
 
@@ -290,15 +282,6 @@ fn parse_args(args: &[String]) -> Result<CliAction, String> {
     }
 }
 
-fn resolve_model_alias(model: &str) -> &str {
-    match model {
-        "opus" => "claude-opus-4-6",
-        "sonnet" => "claude-sonnet-4-6",
-        "haiku" => "claude-haiku-4-5-20251213",
-        _ => model,
-    }
-}
-
 fn normalize_allowed_tools(values: &[String]) -> Result<Option<AllowedToolSet>, String> {
     if values.is_empty() {
         return Ok(None);
@@ -591,19 +574,7 @@ fn print_version() {
 }
 
 fn resume_session(session_path: &Path, commands: &[String]) {
-    let resolved_path = if session_path.exists() {
-        session_path.to_path_buf()
-    } else {
-        match resolve_session_reference(&session_path.display().to_string()) {
-            Ok(handle) => handle.path,
-            Err(error) => {
-                eprintln!("failed to restore session: {error}");
-                std::process::exit(1);
-            }
-        }
-    };
-
-    let session = match Session::load_from_path(&resolved_path) {
+    let session = match Session::load_from_path(session_path) {
         Ok(session) => session,
         Err(error) => {
             eprintln!("failed to restore session: {error}");
@@ -614,7 +585,7 @@ fn resume_session(session_path: &Path, commands: &[String]) {
     if commands.is_empty() {
         println!(
             "Restored session from {} ({} messages).",
-            resolved_path.display(),
+            session_path.display(),
             session.messages.len()
         );
         return;
@@ -626,7 +597,7 @@ fn resume_session(session_path: &Path, commands: &[String]) {
             eprintln!("unsupported resumed command: {raw_command}");
             std::process::exit(2);
         };
-        match run_resume_command(&resolved_path, &session, &command) {
+        match run_resume_command(session_path, &session, &command) {
             Ok(ResumeCommandOutcome {
                 session: next_session,
                 message,
@@ -987,8 +958,6 @@ struct ManagedSessionSummary {
     path: PathBuf,
     modified_epoch_secs: u64,
     message_count: usize,
-    parent_session_id: Option<String>,
-    branch_name: Option<String>,
 }
 
 struct LiveCli {
@@ -996,7 +965,7 @@ struct LiveCli {
     allowed_tools: Option<AllowedToolSet>,
     permission_mode: PermissionMode,
     system_prompt: Vec<String>,
-    runtime: ConversationRuntime<AnthropicRuntimeClient, CliToolExecutor>,
+    runtime: ConversationRuntime<ProviderRuntimeClient, CliToolExecutor>,
     session: SessionHandle,
 }
 
@@ -1008,10 +977,9 @@ impl LiveCli {
         permission_mode: PermissionMode,
     ) -> Result<Self, Box<dyn std::error::Error>> {
         let system_prompt = build_system_prompt()?;
-        let session_state = Session::new();
-        let session = create_managed_session_handle(&session_state.session_id)?;
+        let session = create_managed_session_handle()?;
         let runtime = build_runtime(
-            session_state.with_persistence_path(session.path.clone()),
+            Session::new(),
             model.clone(),
             system_prompt.clone(),
             enable_tools,
@@ -1314,10 +1282,9 @@ impl LiveCli {
             return Ok(false);
         }
 
-        let session_state = Session::new();
-        self.session = create_managed_session_handle(&session_state.session_id)?;
+        self.session = create_managed_session_handle()?;
         self.runtime = build_runtime(
-            session_state.with_persistence_path(self.session.path.clone()),
+            Session::new(),
             self.model.clone(),
             self.system_prompt.clone(),
             true,
@@ -1351,7 +1318,6 @@ impl LiveCli {
         let handle = resolve_session_reference(&session_ref)?;
         let session = Session::load_from_path(&handle.path)?;
         let message_count = session.messages.len();
-        let session_id = session.session_id.clone();
         self.runtime = build_runtime(
             session,
             self.model.clone(),
@@ -1361,10 +1327,7 @@ impl LiveCli {
             self.allowed_tools.clone(),
             self.permission_mode,
         )?;
-        self.session = SessionHandle {
-            id: session_id,
-            path: handle.path,
-        };
+        self.session = handle;
         println!(
             "{}",
             format_resume_report(
@@ -1427,7 +1390,6 @@ impl LiveCli {
                 let handle = resolve_session_reference(target)?;
                 let session = Session::load_from_path(&handle.path)?;
                 let message_count = session.messages.len();
-                let session_id = session.session_id.clone();
                 self.runtime = build_runtime(
                     session,
                     self.model.clone(),
@@ -1437,10 +1399,7 @@ impl LiveCli {
                     self.allowed_tools.clone(),
                     self.permission_mode,
                 )?;
-                self.session = SessionHandle {
-                    id: session_id,
-                    path: handle.path,
-                };
+                self.session = handle;
                 println!(
                     "Session switched\n  Active session   {}\n  File             {}\n  Messages         {}",
                     self.session.id,
@@ -1449,41 +1408,8 @@ impl LiveCli {
                 );
                 Ok(true)
             }
-            Some("fork") => {
-                let forked = self.runtime.fork_session(target.map(ToOwned::to_owned));
-                let parent_session_id = self.session.id.clone();
-                let handle = create_managed_session_handle(&forked.session_id)?;
-                let branch_name = forked
-                    .fork
-                    .as_ref()
-                    .and_then(|fork| fork.branch_name.clone());
-                let forked = forked.with_persistence_path(handle.path.clone());
-                let message_count = forked.messages.len();
-                forked.save_to_path(&handle.path)?;
-                self.runtime = build_runtime(
-                    forked,
-                    self.model.clone(),
-                    self.system_prompt.clone(),
-                    true,
-                    true,
-                    self.allowed_tools.clone(),
-                    self.permission_mode,
-                )?;
-                self.session = handle;
-                println!(
-                    "Session forked\n  Parent session   {}\n  Active session   {}\n  Branch           {}\n  File             {}\n  Messages         {}",
-                    parent_session_id,
-                    self.session.id,
-                    branch_name.as_deref().unwrap_or("(unnamed)"),
-                    self.session.path.display(),
-                    message_count,
-                );
-                Ok(true)
-            }
             Some(other) => {
-                println!(
-                    "Unknown /session action '{other}'. Use /session list, /session switch <session-id>, or /session fork [branch-name]."
-                );
+                println!("Unknown /session action '{other}'. Use /session list or /session switch <session-id>.");
                 Ok(false)
             }
         }
@@ -1516,61 +1442,44 @@ fn sessions_dir() -> Result<PathBuf, Box<dyn std::error::Error>> {
     Ok(path)
 }
 
-fn create_managed_session_handle(
-    session_id: &str,
-) -> Result<SessionHandle, Box<dyn std::error::Error>> {
-    let id = session_id.to_string();
-    let path = sessions_dir()?.join(format!("{id}.{PRIMARY_SESSION_EXTENSION}"));
+fn create_managed_session_handle() -> Result<SessionHandle, Box<dyn std::error::Error>> {
+    let id = generate_session_id();
+    let path = sessions_dir()?.join(format!("{id}.json"));
     Ok(SessionHandle { id, path })
 }
 
+fn generate_session_id() -> String {
+    let millis = SystemTime::now()
+        .duration_since(UNIX_EPOCH)
+        .map(|duration| duration.as_millis())
+        .unwrap_or_default();
+    format!("session-{millis}")
+}
+
 fn resolve_session_reference(reference: &str) -> Result<SessionHandle, Box<dyn std::error::Error>> {
     let direct = PathBuf::from(reference);
-    let looks_like_path = direct.extension().is_some() || direct.components().count() > 1;
     let path = if direct.exists() {
         direct
-    } else if looks_like_path {
-        return Err(format!("session not found: {reference}").into());
     } else {
-        resolve_managed_session_path(reference)?
+        sessions_dir()?.join(format!("{reference}.json"))
     };
+    if !path.exists() {
+        return Err(format!("session not found: {reference}").into());
+    }
     let id = path
-        .file_name()
+        .file_stem()
         .and_then(|value| value.to_str())
-        .and_then(|name| {
-            name.strip_suffix(&format!(".{PRIMARY_SESSION_EXTENSION}"))
-                .or_else(|| name.strip_suffix(&format!(".{LEGACY_SESSION_EXTENSION}")))
-        })
         .unwrap_or(reference)
         .to_string();
     Ok(SessionHandle { id, path })
 }
 
-fn resolve_managed_session_path(session_id: &str) -> Result<PathBuf, Box<dyn std::error::Error>> {
-    let directory = sessions_dir()?;
-    for extension in [PRIMARY_SESSION_EXTENSION, LEGACY_SESSION_EXTENSION] {
-        let path = directory.join(format!("{session_id}.{extension}"));
-        if path.exists() {
-            return Ok(path);
-        }
-    }
-    Err(format!("session not found: {session_id}").into())
-}
-
-fn is_managed_session_file(path: &Path) -> bool {
-    path.extension()
-        .and_then(|ext| ext.to_str())
-        .is_some_and(|extension| {
-            extension == PRIMARY_SESSION_EXTENSION || extension == LEGACY_SESSION_EXTENSION
-        })
-}
-
 fn list_managed_sessions() -> Result<Vec<ManagedSessionSummary>, Box<dyn std::error::Error>> {
     let mut sessions = Vec::new();
     for entry in fs::read_dir(sessions_dir()?)? {
         let entry = entry?;
         let path = entry.path();
-        if !is_managed_session_file(&path) {
+        if path.extension().and_then(|ext| ext.to_str()) != Some("json") {
             continue;
         }
         let metadata = entry.metadata()?;
@@ -1580,41 +1489,19 @@ fn list_managed_sessions() -> Result<Vec<ManagedSessionSummary>, Box<dyn std::er
             .and_then(|time| time.duration_since(UNIX_EPOCH).ok())
             .map(|duration| duration.as_secs())
             .unwrap_or_default();
-        let (id, message_count, parent_session_id, branch_name) = Session::load_from_path(&path)
-            .map(|session| {
-                let parent_session_id = session
-                    .fork
-                    .as_ref()
-                    .map(|fork| fork.parent_session_id.clone());
-                let branch_name = session
-                    .fork
-                    .as_ref()
-                    .and_then(|fork| fork.branch_name.clone());
-                (
-                    session.session_id,
-                    session.messages.len(),
-                    parent_session_id,
-                    branch_name,
-                )
-            })
-            .unwrap_or_else(|_| {
-                (
-                    path.file_stem()
-                        .and_then(|value| value.to_str())
-                        .unwrap_or("unknown")
-                        .to_string(),
-                    0,
-                    None,
-                    None,
-                )
-            });
+        let message_count = Session::load_from_path(&path)
+            .map(|session| session.messages.len())
+            .unwrap_or_default();
+        let id = path
+            .file_stem()
+            .and_then(|value| value.to_str())
+            .unwrap_or("unknown")
+            .to_string();
         sessions.push(ManagedSessionSummary {
             id,
             path,
             modified_epoch_secs,
             message_count,
-            parent_session_id,
-            branch_name,
         });
     }
     sessions.sort_by(|left, right| right.modified_epoch_secs.cmp(&left.modified_epoch_secs));
@@ -1637,23 +1524,11 @@ fn render_session_list(active_session_id: &str) -> Result<String, Box<dyn std::e
         } else {
             "○ saved"
         };
-        let lineage = match (
-            session.branch_name.as_deref(),
-            session.parent_session_id.as_deref(),
-        ) {
-            (Some(branch_name), Some(parent_session_id)) => {
-                format!(" branch={branch_name} from={parent_session_id}")
-            }
-            (None, Some(parent_session_id)) => format!(" from={parent_session_id}"),
-            (Some(branch_name), None) => format!(" branch={branch_name}"),
-            (None, None) => String::new(),
-        };
         lines.push(format!(
-            "  {id:<20} {marker:<10} msgs={msgs:<4} modified={modified}{lineage} path={path}",
+            "  {id:<20} {marker:<10} msgs={msgs:<4} modified={modified} path={path}",
             id = session.id,
             msgs = session.message_count,
             modified = session.modified_epoch_secs,
-            lineage = lineage,
             path = session.path.display(),
         ));
     }
@@ -2030,15 +1905,16 @@ fn build_runtime(
     emit_output: bool,
     allowed_tools: Option<AllowedToolSet>,
     permission_mode: PermissionMode,
-) -> Result<ConversationRuntime<AnthropicRuntimeClient, CliToolExecutor>, Box<dyn std::error::Error>>
+) -> Result<ConversationRuntime<ProviderRuntimeClient, CliToolExecutor>, Box<dyn std::error::Error>>
 {
+    let feature_config = build_runtime_feature_config()?;
     Ok(ConversationRuntime::new_with_features(
         session,
-        AnthropicRuntimeClient::new(model, enable_tools, emit_output, allowed_tools.clone())?,
+        ProviderRuntimeClient::new(model, enable_tools, emit_output, allowed_tools.clone())?,
         CliToolExecutor::new(allowed_tools, emit_output),
         permission_policy(permission_mode),
         system_prompt,
-        build_runtime_feature_config()?,
+        &feature_config,
     ))
 }
 
@@ -2088,26 +1964,33 @@ impl runtime::PermissionPrompter for CliPermissionPrompter {
     }
 }
 
-struct AnthropicRuntimeClient {
+struct ProviderRuntimeClient {
     runtime: tokio::runtime::Runtime,
-    client: AnthropicClient,
+    client: ProviderClient,
     model: String,
     enable_tools: bool,
     emit_output: bool,
     allowed_tools: Option<AllowedToolSet>,
 }
 
-impl AnthropicRuntimeClient {
+impl ProviderRuntimeClient {
     fn new(
         model: String,
         enable_tools: bool,
         emit_output: bool,
         allowed_tools: Option<AllowedToolSet>,
     ) -> Result<Self, Box<dyn std::error::Error>> {
+        let model = resolve_model_alias(&model).to_string();
+        let client = match detect_provider_kind(&model) {
+            ProviderKind::Anthropic => ProviderClient::from_model_with_anthropic_auth(
+                &model,
+                Some(resolve_cli_auth_source()?),
+            )?,
+            ProviderKind::Xai | ProviderKind::OpenAi => ProviderClient::from_model(&model)?,
+        };
         Ok(Self {
             runtime: tokio::runtime::Runtime::new()?,
-            client: AnthropicClient::from_auth(resolve_cli_auth_source()?)
-                .with_base_url(api::read_base_url()),
+            client,
             model,
             enable_tools,
             emit_output,
@@ -2126,7 +2009,7 @@ fn resolve_cli_auth_source() -> Result<AuthSource, Box<dyn std::error::Error>> {
     })?)
 }
 
-impl ApiClient for AnthropicRuntimeClient {
+impl ApiClient for ProviderRuntimeClient {
     #[allow(clippy::too_many_lines)]
     fn stream(&mut self, request: ApiRequest) -> Result<Vec<AssistantEvent>, RuntimeError> {
         let message_request = MessageRequest {
@@ -2164,7 +2047,7 @@ impl ApiClient for AnthropicRuntimeClient {
             let renderer = TerminalRenderer::new();
             let mut markdown_stream = MarkdownStreamState::default();
             let mut events = Vec::new();
-            let mut pending_tool: Option<(String, String, String)> = None;
+            let mut pending_tools: BTreeMap<u32, (String, String, String)> = BTreeMap::new();
             let mut saw_stop = false;
 
             while let Some(event) = stream
@@ -2175,15 +2058,23 @@ impl ApiClient for AnthropicRuntimeClient {
                 match event {
                     ApiStreamEvent::MessageStart(start) => {
                         for block in start.message.content {
-                            push_output_block(block, out, &mut events, &mut pending_tool, true)?;
+                            push_output_block(
+                                block,
+                                0,
+                                out,
+                                &mut events,
+                                &mut pending_tools,
+                                true,
+                            )?;
                         }
                     }
                     ApiStreamEvent::ContentBlockStart(start) => {
                         push_output_block(
                             start.content_block,
+                            start.index,
                             out,
                             &mut events,
-                            &mut pending_tool,
+                            &mut pending_tools,
                             true,
                         )?;
                     }
@@ -2199,18 +2090,18 @@ impl ApiClient for AnthropicRuntimeClient {
                             }
                         }
                         ContentBlockDelta::InputJsonDelta { partial_json } => {
-                            if let Some((_, _, input)) = &mut pending_tool {
+                            if let Some((_, _, input)) = pending_tools.get_mut(&delta.index) {
                                 input.push_str(&partial_json);
                             }
                         }
                     },
-                    ApiStreamEvent::ContentBlockStop(_) => {
+                    ApiStreamEvent::ContentBlockStop(stop) => {
                         if let Some(rendered) = markdown_stream.flush(&renderer) {
                             write!(out, "{rendered}")
                                 .and_then(|()| out.flush())
                                 .map_err(|error| RuntimeError::new(error.to_string()))?;
                         }
-                        if let Some((id, name, input)) = pending_tool.take() {
+                        if let Some((id, name, input)) = pending_tools.remove(&stop.index) {
                             // Display tool call now that input is fully accumulated
                             writeln!(out, "\n{}", format_tool_call_start(&name, &input))
                                 .and_then(|()| out.flush())
@@ -2674,9 +2565,10 @@ fn truncate_for_summary(value: &str, limit: usize) -> String {
 
 fn push_output_block(
     block: OutputContentBlock,
+    block_index: u32,
     out: &mut (impl Write + ?Sized),
     events: &mut Vec<AssistantEvent>,
-    pending_tool: &mut Option<(String, String, String)>,
+    pending_tools: &mut BTreeMap<u32, (String, String, String)>,
     streaming_tool_input: bool,
 ) -> Result<(), RuntimeError> {
     match block {
@@ -2701,7 +2593,7 @@ fn push_output_block(
             } else {
                 input.to_string()
             };
-            *pending_tool = Some((id, name, initial_input));
+            pending_tools.insert(block_index, (id, name, initial_input));
         }
     }
     Ok(())
@@ -2712,11 +2604,13 @@ fn response_to_events(
     out: &mut (impl Write + ?Sized),
 ) -> Result<Vec<AssistantEvent>, RuntimeError> {
     let mut events = Vec::new();
-    let mut pending_tool = None;
+    let mut pending_tools = BTreeMap::new();
 
-    for block in response.content {
-        push_output_block(block, out, &mut events, &mut pending_tool, false)?;
-        if let Some((id, name, input)) = pending_tool.take() {
+    for (index, block) in response.content.into_iter().enumerate() {
+        let index =
+            u32::try_from(index).map_err(|_| RuntimeError::new("response block index overflow"))?;
+        push_output_block(block, index, out, &mut events, &mut pending_tools, false)?;
+        if let Some((id, name, input)) = pending_tools.remove(&index) {
             events.push(AssistantEvent::ToolUse { id, name, input });
         }
     }
@@ -2857,7 +2751,7 @@ fn print_help_to(out: &mut impl Write) -> io::Result<()> {
     writeln!(out, "      Shorthand non-interactive prompt mode")?;
     writeln!(
         out,
-        "  claw --resume SESSION.jsonl [/status] [/compact] [...]"
+        "  claw --resume SESSION.json [/status] [/compact] [...]"
     )?;
     writeln!(
         out,
@@ -2917,7 +2811,7 @@ fn print_help_to(out: &mut impl Write) -> io::Result<()> {
     )?;
     writeln!(
         out,
-        "  claw --resume session.jsonl /status /diff /export notes.txt"
+        "  claw --resume session.json /status /diff /export notes.txt"
     )?;
     writeln!(out, "  claw login")?;
     writeln!(out, "  claw init")?;
@@ -2931,23 +2825,19 @@ fn print_help() {
 #[cfg(test)]
 mod tests {
     use super::{
-        create_managed_session_handle, filter_tool_specs, format_compact_report,
-        format_cost_report, format_model_report, format_model_switch_report,
-        format_permissions_report, format_permissions_switch_report, format_resume_report,
-        format_status_report, format_tool_call_start, format_tool_result,
+        filter_tool_specs, format_compact_report, format_cost_report, format_model_report,
+        format_model_switch_report, format_permissions_report, format_permissions_switch_report,
+        format_resume_report, format_status_report, format_tool_call_start, format_tool_result,
         normalize_permission_mode, parse_args, parse_git_status_metadata, print_help_to,
         push_output_block, render_config_report, render_memory_report, render_repl_help,
-        resolve_model_alias, resolve_session_reference, response_to_events,
-        resume_supported_slash_commands, status_context, CliAction, CliOutputFormat, SlashCommand,
-        StatusUsage, DEFAULT_MODEL,
+        resolve_model_alias, response_to_events, resume_supported_slash_commands, status_context,
+        CliAction, CliOutputFormat, SlashCommand, StatusUsage, DEFAULT_MODEL,
     };
     use api::{MessageResponse, OutputContentBlock, Usage};
-    use runtime::{
-        AssistantEvent, ContentBlock, ConversationMessage, MessageRole, PermissionMode, Session,
-    };
+    use runtime::{AssistantEvent, ContentBlock, ConversationMessage, MessageRole, PermissionMode};
     use serde_json::json;
+    use std::collections::BTreeMap;
     use std::path::PathBuf;
-    use std::sync::{Mutex, OnceLock};
 
     #[test]
     fn defaults_to_repl_when_no_args() {
@@ -3026,6 +2916,9 @@ mod tests {
         assert_eq!(resolve_model_alias("opus"), "claude-opus-4-6");
         assert_eq!(resolve_model_alias("sonnet"), "claude-sonnet-4-6");
         assert_eq!(resolve_model_alias("haiku"), "claude-haiku-4-5-20251213");
+        assert_eq!(resolve_model_alias("grok"), "grok-3");
+        assert_eq!(resolve_model_alias("grok-mini"), "grok-3-mini");
+        assert_eq!(resolve_model_alias("grok-2"), "grok-2");
         assert_eq!(resolve_model_alias("claude-opus"), "claude-opus");
     }
 
@@ -3121,13 +3014,13 @@ mod tests {
     fn parses_resume_flag_with_slash_command() {
         let args = vec![
             "--resume".to_string(),
-            "session.jsonl".to_string(),
+            "session.json".to_string(),
             "/compact".to_string(),
         ];
         assert_eq!(
             parse_args(&args).expect("args should parse"),
             CliAction::ResumeSession {
-                session_path: PathBuf::from("session.jsonl"),
+                session_path: PathBuf::from("session.json"),
                 commands: vec!["/compact".to_string()],
             }
         );
@@ -3137,7 +3030,7 @@ mod tests {
     fn parses_resume_flag_with_multiple_slash_commands() {
         let args = vec![
             "--resume".to_string(),
-            "session.jsonl".to_string(),
+            "session.json".to_string(),
             "/status".to_string(),
             "/compact".to_string(),
             "/cost".to_string(),
@@ -3145,7 +3038,7 @@ mod tests {
         assert_eq!(
             parse_args(&args).expect("args should parse"),
             CliAction::ResumeSession {
-                session_path: PathBuf::from("session.jsonl"),
+                session_path: PathBuf::from("session.json"),
                 commands: vec![
                     "/status".to_string(),
                     "/compact".to_string(),
@@ -3173,7 +3066,7 @@ mod tests {
     fn shared_help_uses_resume_annotation_copy() {
         let help = commands::render_slash_command_help();
         assert!(help.contains("Slash commands"));
-        assert!(help.contains("works with --resume SESSION.jsonl"));
+        assert!(help.contains("works with --resume SESSION.json"));
     }
 
     #[test]
@@ -3193,7 +3086,7 @@ mod tests {
         assert!(help.contains("/diff"));
         assert!(help.contains("/version"));
         assert!(help.contains("/export [file]"));
-        assert!(help.contains("/session [list|switch <session-id>|fork [branch-name]]"));
+        assert!(help.contains("/session [list|switch <session-id>]"));
         assert!(help.contains("/exit"));
     }
 
@@ -3214,9 +3107,9 @@ mod tests {
 
     #[test]
     fn resume_report_uses_sectioned_layout() {
-        let report = format_resume_report("session.jsonl", 14, 6);
+        let report = format_resume_report("session.json", 14, 6);
         assert!(report.contains("Session resumed"));
-        assert!(report.contains("Session file     session.jsonl"));
+        assert!(report.contains("Session file     session.json"));
         assert!(report.contains("Messages         14"));
         assert!(report.contains("Turns            6"));
     }
@@ -3318,7 +3211,7 @@ mod tests {
             "workspace-write",
             &super::StatusContext {
                 cwd: PathBuf::from("/tmp/project"),
-                session_path: Some(PathBuf::from("session.jsonl")),
+                session_path: Some(PathBuf::from("session.json")),
                 loaded_config_files: 2,
                 discovered_config_files: 3,
                 memory_file_count: 4,
@@ -3335,7 +3228,7 @@ mod tests {
         assert!(status.contains("Cwd              /tmp/project"));
         assert!(status.contains("Project root     /tmp"));
         assert!(status.contains("Git branch       main"));
-        assert!(status.contains("Session          session.jsonl"));
+        assert!(status.contains("Session          session.json"));
         assert!(status.contains("Config files     loaded 2/3"));
         assert!(status.contains("Memory files     4"));
     }
@@ -3410,9 +3303,9 @@ mod tests {
     #[test]
     fn parses_resume_and_config_slash_commands() {
         assert_eq!(
-            SlashCommand::parse("/resume saved-session.jsonl"),
+            SlashCommand::parse("/resume saved-session.json"),
             Some(SlashCommand::Resume {
-                session_path: Some("saved-session.jsonl".to_string())
+                session_path: Some("saved-session.json".to_string())
             })
         );
         assert_eq!(
@@ -3431,65 +3324,6 @@ mod tests {
         );
         assert_eq!(SlashCommand::parse("/memory"), Some(SlashCommand::Memory));
         assert_eq!(SlashCommand::parse("/init"), Some(SlashCommand::Init));
-        assert_eq!(
-            SlashCommand::parse("/session fork incident-review"),
-            Some(SlashCommand::Session {
-                action: Some("fork".to_string()),
-                target: Some("incident-review".to_string())
-            })
-        );
-    }
-
-    #[test]
-    fn help_mentions_jsonl_resume_examples() {
-        let mut help = Vec::new();
-        print_help_to(&mut help).expect("help should render");
-        let help = String::from_utf8(help).expect("help should be utf8");
-        assert!(help.contains("claw --resume SESSION.jsonl"));
-        assert!(help.contains("claw --resume session.jsonl /status /diff /export notes.txt"));
-    }
-
-    #[test]
-    fn managed_sessions_default_to_jsonl_and_resolve_legacy_json() {
-        let _guard = cwd_lock().lock().expect("cwd lock");
-        let workspace = temp_workspace("session-resolution");
-        std::fs::create_dir_all(&workspace).expect("workspace should create");
-        let previous = std::env::current_dir().expect("cwd");
-        std::env::set_current_dir(&workspace).expect("switch cwd");
-
-        let handle = create_managed_session_handle("session-alpha").expect("jsonl handle");
-        assert!(handle.path.ends_with("session-alpha.jsonl"));
-
-        let legacy_path = workspace.join(".claude/sessions/legacy.json");
-        std::fs::create_dir_all(
-            legacy_path
-                .parent()
-                .expect("legacy path should have parent directory"),
-        )
-        .expect("session dir should exist");
-        Session::new()
-            .with_persistence_path(legacy_path.clone())
-            .save_to_path(&legacy_path)
-            .expect("legacy session should save");
-
-        let resolved = resolve_session_reference("legacy").expect("legacy session should resolve");
-        assert_eq!(resolved.path, legacy_path);
-
-        std::env::set_current_dir(previous).expect("restore cwd");
-        std::fs::remove_dir_all(workspace).expect("workspace should clean up");
-    }
-
-    fn cwd_lock() -> &'static Mutex<()> {
-        static LOCK: OnceLock<Mutex<()>> = OnceLock::new();
-        LOCK.get_or_init(|| Mutex::new(()))
-    }
-
-    fn temp_workspace(label: &str) -> PathBuf {
-        let nanos = std::time::SystemTime::now()
-            .duration_since(std::time::UNIX_EPOCH)
-            .expect("system time should be after epoch")
-            .as_nanos();
-        std::env::temp_dir().join(format!("claw-cli-{label}-{nanos}"))
     }
 
     #[test]
@@ -3552,15 +3386,16 @@ mod tests {
     fn push_output_block_renders_markdown_text() {
         let mut out = Vec::new();
         let mut events = Vec::new();
-        let mut pending_tool = None;
+        let mut pending_tools = BTreeMap::new();
 
         push_output_block(
             OutputContentBlock::Text {
                 text: "# Heading".to_string(),
             },
+            0,
             &mut out,
             &mut events,
-            &mut pending_tool,
+            &mut pending_tools,
             false,
         )
         .expect("text block should render");
@@ -3574,7 +3409,7 @@ mod tests {
     fn push_output_block_skips_empty_object_prefix_for_tool_streams() {
         let mut out = Vec::new();
         let mut events = Vec::new();
-        let mut pending_tool = None;
+        let mut pending_tools = BTreeMap::new();
 
         push_output_block(
             OutputContentBlock::ToolUse {
@@ -3582,20 +3417,83 @@ mod tests {
                 name: "read_file".to_string(),
                 input: json!({}),
             },
+            1,
             &mut out,
             &mut events,
-            &mut pending_tool,
+            &mut pending_tools,
             true,
         )
         .expect("tool block should accumulate");
 
         assert!(events.is_empty());
         assert_eq!(
-            pending_tool,
+            pending_tools.remove(&1),
             Some(("tool-1".to_string(), "read_file".to_string(), String::new(),))
         );
     }
 
+    #[test]
+    fn pending_tools_preserve_multiple_streaming_tool_calls_by_index() {
+        let mut out = Vec::new();
+        let mut events = Vec::new();
+        let mut pending_tools = BTreeMap::new();
+
+        push_output_block(
+            OutputContentBlock::ToolUse {
+                id: "tool-1".to_string(),
+                name: "read_file".to_string(),
+                input: json!({}),
+            },
+            1,
+            &mut out,
+            &mut events,
+            &mut pending_tools,
+            true,
+        )
+        .expect("first tool should accumulate");
+        push_output_block(
+            OutputContentBlock::ToolUse {
+                id: "tool-2".to_string(),
+                name: "grep_search".to_string(),
+                input: json!({}),
+            },
+            2,
+            &mut out,
+            &mut events,
+            &mut pending_tools,
+            true,
+        )
+        .expect("second tool should accumulate");
+
+        pending_tools
+            .get_mut(&1)
+            .expect("first tool pending")
+            .2
+            .push_str("{\"path\":\"src/main.rs\"}");
+        pending_tools
+            .get_mut(&2)
+            .expect("second tool pending")
+            .2
+            .push_str("{\"pattern\":\"TODO\"}");
+
+        assert_eq!(
+            pending_tools.remove(&1),
+            Some((
+                "tool-1".to_string(),
+                "read_file".to_string(),
+                "{\"path\":\"src/main.rs\"}".to_string(),
+            ))
+        );
+        assert_eq!(
+            pending_tools.remove(&2),
+            Some((
+                "tool-2".to_string(),
+                "grep_search".to_string(),
+                "{\"pattern\":\"TODO\"}".to_string(),
+            ))
+        );
+    }
+
     #[test]
     fn response_to_events_preserves_empty_object_json_input_outside_streaming() {
         let mut out = Vec::new();

rust/crates/tools/src/lib.rs

@@ -4,9 +4,9 @@ use std::process::Command;
 use std::time::{Duration, Instant};
 
 use api::{
-    read_base_url, AnthropicClient, ContentBlockDelta, InputContentBlock, InputMessage,
-    MessageRequest, MessageResponse, OutputContentBlock, StreamEvent as ApiStreamEvent, ToolChoice,
-    ToolDefinition, ToolResultContentBlock,
+    max_tokens_for_model, resolve_model_alias, ContentBlockDelta, InputContentBlock, InputMessage,
+    MessageRequest, MessageResponse, OutputContentBlock, ProviderClient,
+    StreamEvent as ApiStreamEvent, ToolChoice, ToolDefinition, ToolResultContentBlock,
 };
 use reqwest::blocking::Client;
 use runtime::{
@@ -1459,14 +1459,14 @@ fn run_agent_job(job: &AgentJob) -> Result<(), String> {
 
 fn build_agent_runtime(
     job: &AgentJob,
-) -> Result<ConversationRuntime<AnthropicRuntimeClient, SubagentToolExecutor>, String> {
+) -> Result<ConversationRuntime<ProviderRuntimeClient, SubagentToolExecutor>, String> {
     let model = job
         .manifest
         .model
         .clone()
         .unwrap_or_else(|| DEFAULT_AGENT_MODEL.to_string());
     let allowed_tools = job.allowed_tools.clone();
-    let api_client = AnthropicRuntimeClient::new(model, allowed_tools.clone())?;
+    let api_client = ProviderRuntimeClient::new(model, allowed_tools.clone())?;
     let tool_executor = SubagentToolExecutor::new(allowed_tools);
     Ok(ConversationRuntime::new(
         Session::new(),
@@ -1635,18 +1635,17 @@ fn format_agent_terminal_output(status: &str, result: Option<&str>, error: Optio
     sections.join("")
 }
 
-struct AnthropicRuntimeClient {
+struct ProviderRuntimeClient {
     runtime: tokio::runtime::Runtime,
-    client: AnthropicClient,
+    client: ProviderClient,
     model: String,
     allowed_tools: BTreeSet<String>,
 }
 
-impl AnthropicRuntimeClient {
+impl ProviderRuntimeClient {
     fn new(model: String, allowed_tools: BTreeSet<String>) -> Result<Self, String> {
-        let client = AnthropicClient::from_env()
-            .map_err(|error| error.to_string())?
-            .with_base_url(read_base_url());
+        let model = resolve_model_alias(&model).to_string();
+        let client = ProviderClient::from_model(&model).map_err(|error| error.to_string())?;
         Ok(Self {
             runtime: tokio::runtime::Runtime::new().map_err(|error| error.to_string())?,
             client,
@@ -1656,7 +1655,7 @@ impl AnthropicRuntimeClient {
     }
 }
 
-impl ApiClient for AnthropicRuntimeClient {
+impl ApiClient for ProviderRuntimeClient {
     fn stream(&mut self, request: ApiRequest) -> Result<Vec<AssistantEvent>, RuntimeError> {
         let tools = tool_specs_for_allowed_tools(Some(&self.allowed_tools))
             .into_iter()
@@ -1668,7 +1667,7 @@ impl ApiClient for AnthropicRuntimeClient {
             .collect::<Vec<_>>();
         let message_request = MessageRequest {
             model: self.model.clone(),
-            max_tokens: 32_000,
+            max_tokens: max_tokens_for_model(&self.model),
             messages: convert_messages(&request.messages),
             system: (!request.system_prompt.is_empty()).then(|| request.system_prompt.join("\n\n")),
             tools: (!tools.is_empty()).then_some(tools),
@@ -1683,7 +1682,7 @@ impl ApiClient for AnthropicRuntimeClient {
                 .await
                 .map_err(|error| RuntimeError::new(error.to_string()))?;
             let mut events = Vec::new();
-            let mut pending_tool: Option<(String, String, String)> = None;
+            let mut pending_tools: BTreeMap<u32, (String, String, String)> = BTreeMap::new();
             let mut saw_stop = false;
 
             while let Some(event) = stream
@@ -1694,14 +1693,15 @@ impl ApiClient for AnthropicRuntimeClient {
                 match event {
                     ApiStreamEvent::MessageStart(start) => {
                         for block in start.message.content {
-                            push_output_block(block, &mut events, &mut pending_tool, true);
+                            push_output_block(block, 0, &mut events, &mut pending_tools, true);
                         }
                     }
                     ApiStreamEvent::ContentBlockStart(start) => {
                         push_output_block(
                             start.content_block,
+                            start.index,
                             &mut events,
-                            &mut pending_tool,
+                            &mut pending_tools,
                             true,
                         );
                     }
@@ -1712,13 +1712,13 @@ impl ApiClient for AnthropicRuntimeClient {
                             }
                         }
                         ContentBlockDelta::InputJsonDelta { partial_json } => {
-                            if let Some((_, _, input)) = &mut pending_tool {
+                            if let Some((_, _, input)) = pending_tools.get_mut(&delta.index) {
                                 input.push_str(&partial_json);
                             }
                         }
                     },
-                    ApiStreamEvent::ContentBlockStop(_) => {
-                        if let Some((id, name, input)) = pending_tool.take() {
+                    ApiStreamEvent::ContentBlockStop(stop) => {
+                        if let Some((id, name, input)) = pending_tools.remove(&stop.index) {
                             events.push(AssistantEvent::ToolUse { id, name, input });
                         }
                     }
@@ -1839,8 +1839,9 @@ fn convert_messages(messages: &[ConversationMessage]) -> Vec<InputMessage> {
 
 fn push_output_block(
     block: OutputContentBlock,
+    block_index: u32,
     events: &mut Vec<AssistantEvent>,
-    pending_tool: &mut Option<(String, String, String)>,
+    pending_tools: &mut BTreeMap<u32, (String, String, String)>,
     streaming_tool_input: bool,
 ) {
     match block {
@@ -1858,18 +1859,19 @@ fn push_output_block(
             } else {
                 input.to_string()
             };
-            *pending_tool = Some((id, name, initial_input));
+            pending_tools.insert(block_index, (id, name, initial_input));
         }
     }
 }
 
 fn response_to_events(response: MessageResponse) -> Vec<AssistantEvent> {
     let mut events = Vec::new();
-    let mut pending_tool = None;
+    let mut pending_tools = BTreeMap::new();
 
-    for block in response.content {
-        push_output_block(block, &mut events, &mut pending_tool, false);
-        if let Some((id, name, input)) = pending_tool.take() {
+    for (index, block) in response.content.into_iter().enumerate() {
+        let index = u32::try_from(index).expect("response block index overflow");
+        push_output_block(block, index, &mut events, &mut pending_tools, false);
+        if let Some((id, name, input)) = pending_tools.remove(&index) {
             events.push(AssistantEvent::ToolUse { id, name, input });
         }
     }
@@ -2893,6 +2895,7 @@ fn parse_skill_description(contents: &str) -> Option<String> {
 
 #[cfg(test)]
 mod tests {
+    use std::collections::BTreeMap;
     use std::collections::BTreeSet;
     use std::fs;
     use std::io::{Read, Write};
@@ -2905,8 +2908,9 @@ mod tests {
     use super::{
         agent_permission_policy, allowed_tools_for_subagent, execute_agent_with_spawn,
         execute_tool, final_assistant_text, mvp_tool_specs, persist_agent_terminal_state,
-        AgentInput, AgentJob, SubagentToolExecutor,
+        push_output_block, AgentInput, AgentJob, SubagentToolExecutor,
     };
+    use api::OutputContentBlock;
     use runtime::{ApiRequest, AssistantEvent, ConversationRuntime, RuntimeError, Session};
     use serde_json::json;
 
@@ -3121,6 +3125,63 @@ mod tests {
         assert!(error.contains("relative URL without a base") || error.contains("empty host"));
     }
 
+    #[test]
+    fn pending_tools_preserve_multiple_streaming_tool_calls_by_index() {
+        let mut events = Vec::new();
+        let mut pending_tools = BTreeMap::new();
+
+        push_output_block(
+            OutputContentBlock::ToolUse {
+                id: "tool-1".to_string(),
+                name: "read_file".to_string(),
+                input: json!({}),
+            },
+            1,
+            &mut events,
+            &mut pending_tools,
+            true,
+        );
+        push_output_block(
+            OutputContentBlock::ToolUse {
+                id: "tool-2".to_string(),
+                name: "grep_search".to_string(),
+                input: json!({}),
+            },
+            2,
+            &mut events,
+            &mut pending_tools,
+            true,
+        );
+
+        pending_tools
+            .get_mut(&1)
+            .expect("first tool pending")
+            .2
+            .push_str("{\"path\":\"src/main.rs\"}");
+        pending_tools
+            .get_mut(&2)
+            .expect("second tool pending")
+            .2
+            .push_str("{\"pattern\":\"TODO\"}");
+
+        assert_eq!(
+            pending_tools.remove(&1),
+            Some((
+                "tool-1".to_string(),
+                "read_file".to_string(),
+                "{\"path\":\"src/main.rs\"}".to_string(),
+            ))
+        );
+        assert_eq!(
+            pending_tools.remove(&2),
+            Some((
+                "tool-2".to_string(),
+                "grep_search".to_string(),
+                "{\"pattern\":\"TODO\"}".to_string(),
+            ))
+        );
+    }
+
     #[test]
     fn todo_write_persists_and_returns_previous_state() {
         let _guard = env_lock()