feat: telemetry progress

rust/Cargo.lock

@@ -25,6 +25,7 @@ dependencies = [
  "runtime",
  "serde",
  "serde_json",
+ "telemetry",
  "tokio",
 ]
 
@@ -1096,6 +1097,7 @@ dependencies = [
  "serde",
  "serde_json",
  "sha2",
+ "telemetry",
  "tokio",
  "walkdir",
 ]
@@ -1428,6 +1430,14 @@ dependencies = [
  "yaml-rust",
 ]
 
+[[package]]
+name = "telemetry"
+version = "0.1.0"
+dependencies = [
+ "serde",
+ "serde_json",
+]
+
 [[package]]
 name = "thiserror"
 version = "2.0.18"

rust/crates/api/Cargo.toml

@@ -10,6 +10,7 @@ reqwest = { version = "0.12", default-features = false, features = ["json", "rus
 runtime = { path = "../runtime" }
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
+telemetry = { path = "../telemetry" }
 tokio = { version = "1", features = ["io-util", "macros", "net", "rt-multi-thread", "time"] }
 
 [lints]

rust/crates/api/src/error.rs

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

@@ -1,19 +1,13 @@
 mod client;
 mod error;
-mod providers;
 mod sse;
 mod types;
 
 pub use client::{
-    oauth_token_is_expired, read_base_url, read_xai_base_url, resolve_saved_oauth_token,
+    oauth_token_is_expired, read_base_url, resolve_saved_oauth_token, resolve_startup_auth_source,
-    resolve_startup_auth_source, MessageStream, OAuthTokenSet, ProviderClient,
+    AnthropicClient, AuthSource, MessageStream, OAuthTokenSet,
 };
 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,
@@ -21,3 +15,9 @@ pub use types::{
     MessageResponse, MessageStartEvent, MessageStopEvent, OutputContentBlock, StreamEvent,
     ToolChoice, ToolDefinition, ToolResultContentBlock, Usage,
 };
+
+pub use telemetry::{
+    AnalyticsEvent, AnthropicRequestProfile, ClientIdentity, JsonlTelemetrySink,
+    MemoryTelemetrySink, SessionTraceRecord, SessionTracer, TelemetryEvent, TelemetrySink,
+    DEFAULT_ANTHROPIC_VERSION,
+};

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

@@ -1,216 +0,0 @@
-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/src/types.rs

@@ -1,3 +1,4 @@
+use runtime::{pricing_for_model, TokenUsage, UsageCostEstimate};
 use serde::{Deserialize, Serialize};
 use serde_json::Value;
 
@@ -150,7 +151,29 @@ pub struct Usage {
 impl Usage {
     #[must_use]
     pub const fn total_tokens(&self) -> u32 {
-        self.input_tokens + self.output_tokens
+        self.input_tokens
+            + self.output_tokens
+            + self.cache_creation_input_tokens
+            + self.cache_read_input_tokens
+    }
+
+    #[must_use]
+    pub const fn token_usage(&self) -> TokenUsage {
+        TokenUsage {
+            input_tokens: self.input_tokens,
+            output_tokens: self.output_tokens,
+            cache_creation_input_tokens: self.cache_creation_input_tokens,
+            cache_read_input_tokens: self.cache_read_input_tokens,
+        }
+    }
+
+    #[must_use]
+    pub fn estimated_cost_usd(&self, model: &str) -> UsageCostEstimate {
+        let usage = self.token_usage();
+        pricing_for_model(model).map_or_else(
+            || usage.estimate_cost_usd(),
+            |pricing| usage.estimate_cost_usd_with_pricing(pricing),
+        )
     }
 }
 
@@ -210,3 +233,47 @@ pub enum StreamEvent {
     ContentBlockStop(ContentBlockStopEvent),
     MessageStop(MessageStopEvent),
 }
+
+#[cfg(test)]
+mod tests {
+    use runtime::format_usd;
+
+    use super::{MessageResponse, Usage};
+
+    #[test]
+    fn usage_total_tokens_includes_cache_tokens() {
+        let usage = Usage {
+            input_tokens: 10,
+            cache_creation_input_tokens: 2,
+            cache_read_input_tokens: 3,
+            output_tokens: 4,
+        };
+
+        assert_eq!(usage.total_tokens(), 19);
+        assert_eq!(usage.token_usage().total_tokens(), 19);
+    }
+
+    #[test]
+    fn message_response_estimates_cost_from_model_usage() {
+        let response = MessageResponse {
+            id: "msg_cost".to_string(),
+            kind: "message".to_string(),
+            role: "assistant".to_string(),
+            content: Vec::new(),
+            model: "claude-sonnet-4-20250514".to_string(),
+            stop_reason: Some("end_turn".to_string()),
+            stop_sequence: None,
+            usage: Usage {
+                input_tokens: 1_000_000,
+                cache_creation_input_tokens: 100_000,
+                cache_read_input_tokens: 200_000,
+                output_tokens: 500_000,
+            },
+            request_id: None,
+        };
+
+        let cost = response.usage.estimated_cost_usd(&response.model);
+        assert_eq!(format_usd(cost.total_cost_usd()), "$54.6750");
+        assert_eq!(response.total_tokens(), 1_800_000);
+    }
+}

rust/crates/api/tests/client_integration.rs

@@ -3,11 +3,12 @@ use std::sync::Arc;
 use std::time::Duration;
 
 use api::{
-    AnthropicClient, ApiError, AuthSource, ContentBlockDelta, ContentBlockDeltaEvent,
+    AnthropicClient, ApiError, ContentBlockDelta, ContentBlockDeltaEvent, ContentBlockStartEvent,
-    ContentBlockStartEvent, InputContentBlock, InputMessage, MessageDeltaEvent, MessageRequest,
+    InputContentBlock, InputMessage, MessageDeltaEvent, MessageRequest, OutputContentBlock,
-    OutputContentBlock, ProviderClient, StreamEvent, ToolChoice, ToolDefinition,
+    StreamEvent, ToolChoice, ToolDefinition,
 };
 use serde_json::json;
+use telemetry::{ClientIdentity, MemoryTelemetrySink, SessionTracer, TelemetryEvent};
 use tokio::io::{AsyncReadExt, AsyncWriteExt};
 use tokio::net::TcpListener;
 use tokio::sync::Mutex;
@@ -64,6 +65,18 @@ async fn send_message_posts_json_and_parses_response() {
         request.headers.get("authorization").map(String::as_str),
         Some("Bearer proxy-token")
     );
+    assert_eq!(
+        request.headers.get("anthropic-version").map(String::as_str),
+        Some("2023-06-01")
+    );
+    assert_eq!(
+        request.headers.get("user-agent").map(String::as_str),
+        Some("claude-code/0.1.0")
+    );
+    assert_eq!(
+        request.headers.get("anthropic-beta").map(String::as_str),
+        Some("claude-code-20250219,prompt-caching-scope-2026-01-05")
+    );
     let body: serde_json::Value =
         serde_json::from_str(&request.body).expect("request body should be json");
     assert_eq!(
@@ -73,6 +86,115 @@ async fn send_message_posts_json_and_parses_response() {
     assert!(body.get("stream").is_none());
     assert_eq!(body["tools"][0]["name"], json!("get_weather"));
     assert_eq!(body["tool_choice"]["type"], json!("auto"));
+    assert_eq!(
+        body["betas"],
+        json!(["claude-code-20250219", "prompt-caching-scope-2026-01-05"])
+    );
+}
+
+#[tokio::test]
+async fn send_message_applies_request_profile_and_records_telemetry() {
+    let state = Arc::new(Mutex::new(Vec::<CapturedRequest>::new()));
+    let server = spawn_server(
+        state.clone(),
+        vec![http_response_with_headers(
+            "200 OK",
+            "application/json",
+            concat!(
+                "{",
+                "\"id\":\"msg_profile\",",
+                "\"type\":\"message\",",
+                "\"role\":\"assistant\",",
+                "\"content\":[{\"type\":\"text\",\"text\":\"ok\"}],",
+                "\"model\":\"claude-3-7-sonnet-latest\",",
+                "\"stop_reason\":\"end_turn\",",
+                "\"stop_sequence\":null,",
+                "\"usage\":{\"input_tokens\":1,\"cache_creation_input_tokens\":2,\"cache_read_input_tokens\":3,\"output_tokens\":1}",
+                "}"
+            ),
+            &[("request-id", "req_profile_123")],
+        )],
+    )
+    .await;
+    let sink = Arc::new(MemoryTelemetrySink::default());
+
+    let client = AnthropicClient::new("test-key")
+        .with_base_url(server.base_url())
+        .with_client_identity(ClientIdentity::new("claude-code", "9.9.9").with_runtime("rust-cli"))
+        .with_beta("tools-2026-04-01")
+        .with_extra_body_param("metadata", json!({"source": "clawd-code"}))
+        .with_session_tracer(SessionTracer::new("session-telemetry", sink.clone()));
+
+    let response = client
+        .send_message(&sample_request(false))
+        .await
+        .expect("request should succeed");
+
+    assert_eq!(response.request_id.as_deref(), Some("req_profile_123"));
+
+    let captured = state.lock().await;
+    let request = captured.first().expect("server should capture request");
+    assert_eq!(
+        request.headers.get("anthropic-beta").map(String::as_str),
+        Some("claude-code-20250219,prompt-caching-scope-2026-01-05,tools-2026-04-01")
+    );
+    assert_eq!(
+        request.headers.get("user-agent").map(String::as_str),
+        Some("claude-code/9.9.9")
+    );
+    let body: serde_json::Value =
+        serde_json::from_str(&request.body).expect("request body should be json");
+    assert_eq!(body["metadata"]["source"], json!("clawd-code"));
+    assert_eq!(
+        body["betas"],
+        json!([
+            "claude-code-20250219",
+            "prompt-caching-scope-2026-01-05",
+            "tools-2026-04-01"
+        ])
+    );
+
+    let events = sink.events();
+    assert_eq!(events.len(), 6);
+    assert!(matches!(
+        &events[0],
+        TelemetryEvent::HttpRequestStarted {
+            session_id,
+            attempt: 1,
+            method,
+            path,
+            ..
+        } if session_id == "session-telemetry" && method == "POST" && path == "/v1/messages"
+    ));
+    assert!(matches!(
+        &events[1],
+        TelemetryEvent::SessionTrace(trace) if trace.name == "http_request_started"
+    ));
+    assert!(matches!(
+        &events[2],
+        TelemetryEvent::HttpRequestSucceeded {
+            request_id,
+            status: 200,
+            ..
+        } if request_id.as_deref() == Some("req_profile_123")
+    ));
+    assert!(matches!(
+        &events[3],
+        TelemetryEvent::SessionTrace(trace) if trace.name == "http_request_succeeded"
+    ));
+    assert!(matches!(
+        &events[4],
+        TelemetryEvent::Analytics(event)
+            if event.namespace == "api"
+                && event.action == "message_usage"
+                && event.properties.get("request_id") == Some(&json!("req_profile_123"))
+                && event.properties.get("total_tokens") == Some(&json!(7))
+                && event.properties.get("estimated_cost_usd") == Some(&json!("$0.0001"))
+    ));
+    assert!(matches!(
+        &events[5],
+        TelemetryEvent::SessionTrace(trace) if trace.name == "analytics"
+    ));
 }
 
 #[tokio::test]
@@ -195,47 +317,6 @@ 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

@@ -1,415 +0,0 @@
-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

@@ -1,86 +0,0 @@
-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/runtime/Cargo.toml

@@ -11,6 +11,7 @@ glob = "0.3"
 regex = "1"
 serde = { version = "1", features = ["derive"] }
 serde_json = "1"
+telemetry = { path = "../telemetry" }
 tokio = { version = "1", features = ["io-util", "macros", "process", "rt", "rt-multi-thread", "time"] }
 walkdir = "2"
 

rust/crates/runtime/src/conversation.rs

@@ -1,6 +1,9 @@
 use std::collections::BTreeMap;
 use std::fmt::{Display, Formatter};
 
+use serde_json::{Map, Value};
+use telemetry::SessionTracer;
+
 use crate::compact::{
     compact_session, estimate_session_tokens, CompactionConfig, CompactionResult,
 };
@@ -97,6 +100,7 @@ pub struct ConversationRuntime<C, T> {
     max_iterations: usize,
     usage_tracker: UsageTracker,
     hook_runner: HookRunner,
+    session_tracer: Option<SessionTracer>,
 }
 
 impl<C, T> ConversationRuntime<C, T>
@@ -141,6 +145,7 @@ where
             max_iterations: usize::MAX,
             usage_tracker,
             hook_runner: HookRunner::from_feature_config(feature_config),
+            session_tracer: None,
         }
     }
 
@@ -150,14 +155,23 @@ where
         self
     }
 
+    #[must_use]
+    pub fn with_session_tracer(mut self, session_tracer: SessionTracer) -> Self {
+        self.session_tracer = Some(session_tracer);
+        self
+    }
+
+    #[allow(clippy::too_many_lines)]
     pub fn run_turn(
         &mut self,
         user_input: impl Into<String>,
         mut prompter: Option<&mut dyn PermissionPrompter>,
     ) -> Result<TurnSummary, RuntimeError> {
+        let user_input = user_input.into();
+        self.record_turn_started(&user_input);
         self.session
             .messages
-            .push(ConversationMessage::user_text(user_input.into()));
+            .push(ConversationMessage::user_text(user_input));
 
         let mut assistant_messages = Vec::new();
         let mut tool_results = Vec::new();
@@ -166,17 +180,31 @@ where
         loop {
             iterations += 1;
             if iterations > self.max_iterations {
-                return Err(RuntimeError::new(
+                let error = RuntimeError::new(
                     "conversation loop exceeded the maximum number of iterations",
-                ));
+                );
+                self.record_turn_failed(iterations, &error);
+                return Err(error);
             }
 
             let request = ApiRequest {
                 system_prompt: self.system_prompt.clone(),
                 messages: self.session.messages.clone(),
             };
-            let events = self.api_client.stream(request)?;
+            let events = match self.api_client.stream(request) {
-            let (assistant_message, usage) = build_assistant_message(events)?;
+                Ok(events) => events,
+                Err(error) => {
+                    self.record_turn_failed(iterations, &error);
+                    return Err(error);
+                }
+            };
+            let (assistant_message, usage) = match build_assistant_message(events) {
+                Ok(result) => result,
+                Err(error) => {
+                    self.record_turn_failed(iterations, &error);
+                    return Err(error);
+                }
+            };
             if let Some(usage) = usage {
                 self.usage_tracker.record(usage);
             }
@@ -190,6 +218,11 @@ where
                     _ => None,
                 })
                 .collect::<Vec<_>>();
+            self.record_assistant_iteration(
+                iterations,
+                &assistant_message,
+                pending_tool_uses.len(),
+            );
 
             self.session.messages.push(assistant_message.clone());
             assistant_messages.push(assistant_message);
@@ -199,6 +232,7 @@ where
             }
 
             for (tool_use_id, tool_name, input) in pending_tool_uses {
+                self.record_tool_started(iterations, &tool_name);
                 let permission_outcome = if let Some(prompt) = prompter.as_mut() {
                     self.permission_policy
                         .authorize(&tool_name, &input, Some(*prompt))
@@ -249,17 +283,20 @@ where
                         ConversationMessage::tool_result(tool_use_id, tool_name, reason, true)
                     }
                 };
+                self.record_tool_finished(iterations, &result_message);
                 self.session.messages.push(result_message.clone());
                 tool_results.push(result_message);
             }
         }
 
-        Ok(TurnSummary {
+        let summary = TurnSummary {
             assistant_messages,
             tool_results,
             iterations,
             usage: self.usage_tracker.cumulative_usage(),
-        })
+        };
+        self.record_turn_completed(&summary);
+        Ok(summary)
     }
 
     #[must_use]
@@ -286,6 +323,126 @@ where
     pub fn into_session(self) -> Session {
         self.session
     }
+
+    fn record_turn_started(&self, user_input: &str) {
+        if let Some(tracer) = &self.session_tracer {
+            let mut attributes = Map::new();
+            attributes.insert(
+                "message_count_before".to_string(),
+                Value::from(u64::try_from(self.session.messages.len()).unwrap_or(u64::MAX)),
+            );
+            attributes.insert(
+                "input_chars".to_string(),
+                Value::from(u64::try_from(user_input.chars().count()).unwrap_or(u64::MAX)),
+            );
+            tracer.record("turn_started", attributes);
+        }
+    }
+
+    fn record_assistant_iteration(
+        &self,
+        iteration: usize,
+        assistant_message: &ConversationMessage,
+        pending_tool_count: usize,
+    ) {
+        if let Some(tracer) = &self.session_tracer {
+            let mut attributes = Map::new();
+            attributes.insert(
+                "iteration".to_string(),
+                Value::from(u64::try_from(iteration).unwrap_or(u64::MAX)),
+            );
+            attributes.insert(
+                "block_count".to_string(),
+                Value::from(u64::try_from(assistant_message.blocks.len()).unwrap_or(u64::MAX)),
+            );
+            attributes.insert(
+                "pending_tool_count".to_string(),
+                Value::from(u64::try_from(pending_tool_count).unwrap_or(u64::MAX)),
+            );
+            tracer.record("assistant_iteration_completed", attributes);
+        }
+    }
+
+    fn record_tool_started(&self, iteration: usize, tool_name: &str) {
+        if let Some(tracer) = &self.session_tracer {
+            let mut attributes = Map::new();
+            attributes.insert(
+                "iteration".to_string(),
+                Value::from(u64::try_from(iteration).unwrap_or(u64::MAX)),
+            );
+            attributes.insert(
+                "tool_name".to_string(),
+                Value::String(tool_name.to_string()),
+            );
+            tracer.record("tool_execution_started", attributes);
+        }
+    }
+
+    fn record_tool_finished(&self, iteration: usize, result_message: &ConversationMessage) {
+        let Some(tracer) = &self.session_tracer else {
+            return;
+        };
+        let Some(ContentBlock::ToolResult {
+            tool_name,
+            is_error,
+            output,
+            ..
+        }) = result_message.blocks.first()
+        else {
+            return;
+        };
+        let mut attributes = Map::new();
+        attributes.insert(
+            "iteration".to_string(),
+            Value::from(u64::try_from(iteration).unwrap_or(u64::MAX)),
+        );
+        attributes.insert("tool_name".to_string(), Value::String(tool_name.clone()));
+        attributes.insert("is_error".to_string(), Value::Bool(*is_error));
+        attributes.insert(
+            "output_chars".to_string(),
+            Value::from(u64::try_from(output.chars().count()).unwrap_or(u64::MAX)),
+        );
+        tracer.record("tool_execution_finished", attributes);
+    }
+
+    fn record_turn_completed(&self, summary: &TurnSummary) {
+        if let Some(tracer) = &self.session_tracer {
+            let mut attributes = Map::new();
+            attributes.insert(
+                "assistant_message_count".to_string(),
+                Value::from(u64::try_from(summary.assistant_messages.len()).unwrap_or(u64::MAX)),
+            );
+            attributes.insert(
+                "tool_result_count".to_string(),
+                Value::from(u64::try_from(summary.tool_results.len()).unwrap_or(u64::MAX)),
+            );
+            attributes.insert(
+                "iterations".to_string(),
+                Value::from(u64::try_from(summary.iterations).unwrap_or(u64::MAX)),
+            );
+            attributes.insert(
+                "total_input_tokens".to_string(),
+                Value::from(summary.usage.input_tokens),
+            );
+            attributes.insert(
+                "total_output_tokens".to_string(),
+                Value::from(summary.usage.output_tokens),
+            );
+            tracer.record("turn_completed", attributes);
+        }
+    }
+
+    fn record_turn_failed(&self, iteration: usize, error: &RuntimeError) {
+        if let Some(tracer) = &self.session_tracer {
+            let mut attributes = Map::new();
+            attributes.insert(
+                "iteration".to_string(),
+                Value::from(u64::try_from(iteration).unwrap_or(u64::MAX)),
+            );
+            attributes.insert("error".to_string(), Value::String(error.to_string()));
+            tracer.record("turn_failed", attributes);
+        }
+    }
 }
 
 fn build_assistant_message(
@@ -409,6 +566,8 @@ mod tests {
     use crate::session::{ContentBlock, MessageRole, Session};
     use crate::usage::TokenUsage;
     use std::path::PathBuf;
+    use std::sync::Arc;
+    use telemetry::{MemoryTelemetrySink, SessionTracer, TelemetryEvent};
 
     struct ScriptedApiClient {
         call_count: usize,
@@ -521,6 +680,39 @@ mod tests {
         ));
     }
 
+    #[test]
+    fn records_runtime_session_trace_events() {
+        let sink = Arc::new(MemoryTelemetrySink::default());
+        let tracer = SessionTracer::new("session-runtime", sink.clone());
+        let mut runtime = ConversationRuntime::new(
+            Session::new(),
+            ScriptedApiClient { call_count: 0 },
+            StaticToolExecutor::new().register("add", |_input| Ok("4".to_string())),
+            PermissionPolicy::new(PermissionMode::WorkspaceWrite),
+            vec!["system".to_string()],
+        )
+        .with_session_tracer(tracer);
+
+        runtime
+            .run_turn("what is 2 + 2?", Some(&mut PromptAllowOnce))
+            .expect("conversation loop should succeed");
+
+        let events = sink.events();
+        let trace_names = events
+            .iter()
+            .filter_map(|event| match event {
+                TelemetryEvent::SessionTrace(trace) => Some(trace.name.as_str()),
+                _ => None,
+            })
+            .collect::<Vec<_>>();
+
+        assert!(trace_names.contains(&"turn_started"));
+        assert!(trace_names.contains(&"assistant_iteration_completed"));
+        assert!(trace_names.contains(&"tool_execution_started"));
+        assert!(trace_names.contains(&"tool_execution_finished"));
+        assert!(trace_names.contains(&"turn_completed"));
+    }
+
     #[test]
     fn records_denied_tool_results_when_prompt_rejects() {
         struct RejectPrompter;

rust/crates/runtime/src/hooks.rs

@@ -51,16 +51,6 @@ 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 {
@@ -74,7 +64,7 @@ impl HookRunner {
 
     #[must_use]
     pub fn run_pre_tool_use(&self, tool_name: &str, tool_input: &str) -> HookRunResult {
-        self.run_commands(
+        Self::run_commands(
             HookEvent::PreToolUse,
             self.config.pre_tool_use(),
             tool_name,
@@ -92,7 +82,7 @@ impl HookRunner {
         tool_output: &str,
         is_error: bool,
     ) -> HookRunResult {
-        self.run_commands(
+        Self::run_commands(
             HookEvent::PostToolUse,
             self.config.post_tool_use(),
             tool_name,
@@ -103,7 +93,6 @@ impl HookRunner {
     }
 
     fn run_commands(
-        &self,
         event: HookEvent,
         commands: &[String],
         tool_name: &str,
@@ -124,21 +113,19 @@ impl HookRunner {
             "tool_result_is_error": is_error,
         })
         .to_string();
+        let invocation = HookInvocation {
+            event,
+            tool_name,
+            tool_input,
+            tool_output,
+            is_error,
+            payload: &payload,
+        };
 
         let mut messages = Vec::new();
 
         for command in commands {
-            match Self::run_command(
+            match Self::run_command(command, &invocation) {
-                command,
-                HookCommandRequest {
-                    event,
-                    tool_name,
-                    tool_input,
-                    tool_output,
-                    is_error,
-                    payload: &payload,
-                },
-            ) {
                 HookCommandOutcome::Allow { message } => {
                     if let Some(message) = message {
                         messages.push(message);
@@ -161,23 +148,23 @@ impl HookRunner {
         HookRunResult::allow(messages)
     }
 
-    fn run_command(command: &str, request: HookCommandRequest<'_>) -> HookCommandOutcome {
+    fn run_command(command: &str, invocation: &HookInvocation<'_>) -> 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", request.event.as_str());
+        child.env("HOOK_EVENT", invocation.event.as_str());
-        child.env("HOOK_TOOL_NAME", request.tool_name);
+        child.env("HOOK_TOOL_NAME", invocation.tool_name);
-        child.env("HOOK_TOOL_INPUT", request.tool_input);
+        child.env("HOOK_TOOL_INPUT", invocation.tool_input);
         child.env(
             "HOOK_TOOL_IS_ERROR",
-            if request.is_error { "1" } else { "0" },
+            if invocation.is_error { "1" } else { "0" },
         );
-        if let Some(tool_output) = request.tool_output {
+        if let Some(tool_output) = invocation.tool_output {
             child.env("HOOK_TOOL_OUTPUT", tool_output);
         }
 
-        match child.output_with_stdin(request.payload.as_bytes()) {
+        match child.output_with_stdin(invocation.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();
@@ -196,23 +183,32 @@ impl HookRunner {
                     None => HookCommandOutcome::Warn {
                         message: format!(
                             "{} hook `{command}` terminated by signal while handling `{}`",
-                            request.event.as_str(),
+                            invocation.event.as_str(),
-                            request.tool_name
+                            invocation.tool_name
                         ),
                     },
                 }
             }
             Err(error) => HookCommandOutcome::Warn {
                 message: format!(
-                    "{} hook `{command}` failed to start for `{}`: {error}",
+                    "{} hook `{command}` failed to start for `{tool_name}`: {error}",
-                    request.event.as_str(),
+                    invocation.event.as_str(),
-                    request.tool_name
+                    tool_name = invocation.tool_name
                 ),
             },
         }
     }
 }
 
+struct HookInvocation<'a> {
+    event: HookEvent,
+    tool_name: &'a str,
+    tool_input: &'a str,
+    tool_output: Option<&'a str>,
+    is_error: bool,
+    payload: &'a str,
+}
+
 enum HookCommandOutcome {
     Allow { message: Option<String> },
     Deny { message: Option<String> },

rust/crates/runtime/src/mcp_stdio.rs

@@ -1144,8 +1144,20 @@ mod tests {
     }
 
     fn cleanup_script(script_path: &Path) {
-        fs::remove_file(script_path).expect("cleanup script");
+        if let Err(error) = fs::remove_file(script_path) {
-        fs::remove_dir_all(script_path.parent().expect("script parent")).expect("cleanup dir");
+            assert_eq!(
+                error.kind(),
+                std::io::ErrorKind::NotFound,
+                "cleanup script: {error}"
+            );
+        }
+        if let Err(error) = fs::remove_dir_all(script_path.parent().expect("script parent")) {
+            assert_eq!(
+                error.kind(),
+                std::io::ErrorKind::NotFound,
+                "cleanup dir: {error}"
+            );
+        }
     }
 
     fn manager_server_config(

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

@@ -4,6 +4,7 @@ mod render;
 
 use std::collections::{BTreeMap, BTreeSet};
 use std::env;
+use std::fmt::Write as _;
 use std::fs;
 use std::io::{self, Read, Write};
 use std::net::TcpListener;
@@ -12,10 +13,10 @@ use std::process::Command;
 use std::time::{SystemTime, UNIX_EPOCH};
 
 use api::{
-    detect_provider_kind, max_tokens_for_model, resolve_model_alias, resolve_startup_auth_source,
+    resolve_startup_auth_source, AnthropicClient, AuthSource, ContentBlockDelta, InputContentBlock,
-    AnthropicClient, AuthSource, ContentBlockDelta, InputContentBlock, InputMessage,
+    InputMessage, JsonlTelemetrySink, MessageRequest, MessageResponse, OutputContentBlock,
-    MessageRequest, MessageResponse, OutputContentBlock, ProviderClient, ProviderKind,
+    SessionTracer, StreamEvent as ApiStreamEvent, ToolChoice, ToolDefinition,
-    StreamEvent as ApiStreamEvent, ToolChoice, ToolDefinition, ToolResultContentBlock,
+    ToolResultContentBlock,
 };
 
 use commands::{
@@ -36,8 +37,16 @@ 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 TELEMETRY_LOG_PATH_ENV: &str = "CLAW_TELEMETRY_LOG_PATH";
 const VERSION: &str = env!("CARGO_PKG_VERSION");
 const BUILD_TARGET: Option<&str> = option_env!("TARGET");
 const GIT_SHA: Option<&str> = option_env!("GIT_SHA");
@@ -282,6 +291,15 @@ 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);
@@ -965,7 +983,7 @@ struct LiveCli {
     allowed_tools: Option<AllowedToolSet>,
     permission_mode: PermissionMode,
     system_prompt: Vec<String>,
-    runtime: ConversationRuntime<ProviderRuntimeClient, CliToolExecutor>,
+    runtime: ConversationRuntime<AnthropicRuntimeClient, CliToolExecutor>,
     session: SessionHandle,
 }
 
@@ -980,6 +998,7 @@ impl LiveCli {
         let session = create_managed_session_handle()?;
         let runtime = build_runtime(
             Session::new(),
+            &session.id,
             model.clone(),
             system_prompt.clone(),
             enable_tools,
@@ -1071,6 +1090,7 @@ impl LiveCli {
         let session = self.runtime.session().clone();
         let mut runtime = build_runtime(
             session,
+            &self.session.id,
             self.model.clone(),
             self.system_prompt.clone(),
             true,
@@ -1217,6 +1237,7 @@ impl LiveCli {
         let message_count = session.messages.len();
         self.runtime = build_runtime(
             session,
+            &self.session.id,
             model.clone(),
             self.system_prompt.clone(),
             true,
@@ -1260,6 +1281,7 @@ impl LiveCli {
         self.permission_mode = permission_mode_from_label(normalized);
         self.runtime = build_runtime(
             session,
+            &self.session.id,
             self.model.clone(),
             self.system_prompt.clone(),
             true,
@@ -1285,6 +1307,7 @@ impl LiveCli {
         self.session = create_managed_session_handle()?;
         self.runtime = build_runtime(
             Session::new(),
+            &self.session.id,
             self.model.clone(),
             self.system_prompt.clone(),
             true,
@@ -1320,6 +1343,7 @@ impl LiveCli {
         let message_count = session.messages.len();
         self.runtime = build_runtime(
             session,
+            &handle.id,
             self.model.clone(),
             self.system_prompt.clone(),
             true,
@@ -1392,6 +1416,7 @@ impl LiveCli {
                 let message_count = session.messages.len();
                 self.runtime = build_runtime(
                     session,
+                    &handle.id,
                     self.model.clone(),
                     self.system_prompt.clone(),
                     true,
@@ -1422,6 +1447,7 @@ impl LiveCli {
         let skipped = removed == 0;
         self.runtime = build_runtime(
             result.compacted_session,
+            &self.session.id,
             self.model.clone(),
             self.system_prompt.clone(),
             true,
@@ -1897,25 +1923,53 @@ fn build_runtime_feature_config(
         .clone())
 }
 
+#[allow(clippy::too_many_arguments)]
 fn build_runtime(
     session: Session,
+    session_id: &str,
     model: String,
     system_prompt: Vec<String>,
     enable_tools: bool,
     emit_output: bool,
     allowed_tools: Option<AllowedToolSet>,
     permission_mode: PermissionMode,
-) -> Result<ConversationRuntime<ProviderRuntimeClient, CliToolExecutor>, Box<dyn std::error::Error>>
+) -> Result<ConversationRuntime<AnthropicRuntimeClient, CliToolExecutor>, Box<dyn std::error::Error>>
 {
-    let feature_config = build_runtime_feature_config()?;
+    let session_tracer = build_session_tracer(session_id)?;
-    Ok(ConversationRuntime::new_with_features(
+    let api_client = match session_tracer.clone() {
+        Some(session_tracer) => {
+            AnthropicRuntimeClient::new(model, enable_tools, emit_output, allowed_tools.clone())?
+                .with_session_tracer(session_tracer)
+        }
+        None => {
+            AnthropicRuntimeClient::new(model, enable_tools, emit_output, allowed_tools.clone())?
+        }
+    };
+    let runtime = ConversationRuntime::new_with_features(
         session,
-        ProviderRuntimeClient::new(model, enable_tools, emit_output, allowed_tools.clone())?,
+        api_client,
         CliToolExecutor::new(allowed_tools, emit_output),
         permission_policy(permission_mode),
         system_prompt,
-        &feature_config,
+        &build_runtime_feature_config()?,
-    ))
+    );
+    Ok(match session_tracer {
+        Some(session_tracer) => runtime.with_session_tracer(session_tracer),
+        None => runtime,
+    })
+}
+
+fn build_session_tracer(
+    session_id: &str,
+) -> Result<Option<SessionTracer>, Box<dyn std::error::Error>> {
+    let Some(path) = env::var_os(TELEMETRY_LOG_PATH_ENV) else {
+        return Ok(None);
+    };
+    let sink = JsonlTelemetrySink::new(PathBuf::from(path))?;
+    Ok(Some(SessionTracer::new(
+        session_id.to_string(),
+        std::sync::Arc::new(sink),
+    )))
 }
 
 struct CliPermissionPrompter {
@@ -1964,39 +2018,37 @@ impl runtime::PermissionPrompter for CliPermissionPrompter {
     }
 }
 
-struct ProviderRuntimeClient {
+struct AnthropicRuntimeClient {
     runtime: tokio::runtime::Runtime,
-    client: ProviderClient,
+    client: AnthropicClient,
     model: String,
     enable_tools: bool,
     emit_output: bool,
     allowed_tools: Option<AllowedToolSet>,
 }
 
-impl ProviderRuntimeClient {
+impl AnthropicRuntimeClient {
     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,
+            client: AnthropicClient::from_auth(resolve_cli_auth_source()?)
+                .with_base_url(api::read_base_url()),
             model,
             enable_tools,
             emit_output,
             allowed_tools,
         })
     }
+
+    fn with_session_tracer(mut self, session_tracer: SessionTracer) -> Self {
+        self.client = self.client.with_session_tracer(session_tracer);
+        self
+    }
 }
 
 fn resolve_cli_auth_source() -> Result<AuthSource, Box<dyn std::error::Error>> {
@@ -2009,7 +2061,7 @@ fn resolve_cli_auth_source() -> Result<AuthSource, Box<dyn std::error::Error>> {
     })?)
 }
 
-impl ApiClient for ProviderRuntimeClient {
+impl ApiClient for AnthropicRuntimeClient {
     #[allow(clippy::too_many_lines)]
     fn stream(&mut self, request: ApiRequest) -> Result<Vec<AssistantEvent>, RuntimeError> {
         let message_request = MessageRequest {
@@ -2047,7 +2099,7 @@ impl ApiClient for ProviderRuntimeClient {
             let renderer = TerminalRenderer::new();
             let mut markdown_stream = MarkdownStreamState::default();
             let mut events = Vec::new();
-            let mut pending_tools: BTreeMap<u32, (String, String, String)> = BTreeMap::new();
+            let mut pending_tool: Option<(String, String, String)> = None;
             let mut saw_stop = false;
 
             while let Some(event) = stream
@@ -2058,23 +2110,15 @@ impl ApiClient for ProviderRuntimeClient {
                 match event {
                     ApiStreamEvent::MessageStart(start) => {
                         for block in start.message.content {
-                            push_output_block(
+                            push_output_block(block, out, &mut events, &mut pending_tool, true)?;
-                                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_tools,
+                            &mut pending_tool,
                             true,
                         )?;
                     }
@@ -2090,18 +2134,18 @@ impl ApiClient for ProviderRuntimeClient {
                             }
                         }
                         ContentBlockDelta::InputJsonDelta { partial_json } => {
-                            if let Some((_, _, input)) = pending_tools.get_mut(&delta.index) {
+                            if let Some((_, _, input)) = &mut pending_tool {
                                 input.push_str(&partial_json);
                             }
                         }
                     },
-                    ApiStreamEvent::ContentBlockStop(stop) => {
+                    ApiStreamEvent::ContentBlockStop(_) => {
                         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_tools.remove(&stop.index) {
+                        if let Some((id, name, input)) = pending_tool.take() {
                             // Display tool call now that input is fully accumulated
                             writeln!(out, "\n{}", format_tool_call_start(&name, &input))
                                 .and_then(|()| out.flush())
@@ -2110,12 +2154,7 @@ impl ApiClient for ProviderRuntimeClient {
                         }
                     }
                     ApiStreamEvent::MessageDelta(delta) => {
-                        events.push(AssistantEvent::Usage(TokenUsage {
+                        events.push(AssistantEvent::Usage(delta.usage.token_usage()));
-                            input_tokens: delta.usage.input_tokens,
-                            output_tokens: delta.usage.output_tokens,
-                            cache_creation_input_tokens: 0,
-                            cache_read_input_tokens: 0,
-                        }));
                     }
                     ApiStreamEvent::MessageStop(_) => {
                         saw_stop = true;
@@ -2365,13 +2404,13 @@ fn format_bash_result(icon: &str, parsed: &serde_json::Value) -> String {
         .get("backgroundTaskId")
         .and_then(|value| value.as_str())
     {
-        lines[0].push_str(&format!(" backgrounded ({task_id})"));
+        write!(&mut lines[0], " backgrounded ({task_id})").expect("write to string");
     } else if let Some(status) = parsed
         .get("returnCodeInterpretation")
         .and_then(|value| value.as_str())
         .filter(|status| !status.is_empty())
     {
-        lines[0].push_str(&format!(" {status}"));
+        write!(&mut lines[0], " {status}").expect("write to string");
     }
 
     if let Some(stdout) = parsed.get("stdout").and_then(|value| value.as_str()) {
@@ -2393,15 +2432,15 @@ fn format_read_result(icon: &str, parsed: &serde_json::Value) -> String {
     let path = extract_tool_path(file);
     let start_line = file
         .get("startLine")
-        .and_then(|value| value.as_u64())
+        .and_then(serde_json::Value::as_u64)
         .unwrap_or(1);
     let num_lines = file
         .get("numLines")
-        .and_then(|value| value.as_u64())
+        .and_then(serde_json::Value::as_u64)
         .unwrap_or(0);
     let total_lines = file
         .get("totalLines")
-        .and_then(|value| value.as_u64())
+        .and_then(serde_json::Value::as_u64)
         .unwrap_or(num_lines);
     let content = file
         .get("content")
@@ -2427,8 +2466,7 @@ fn format_write_result(icon: &str, parsed: &serde_json::Value) -> String {
     let line_count = parsed
         .get("content")
         .and_then(|value| value.as_str())
-        .map(|content| content.lines().count())
+        .map_or(0, |content| content.lines().count());
-        .unwrap_or(0);
     format!(
         "{icon} \x1b[1;32m✏️ {} {path}\x1b[0m \x1b[2m({line_count} lines)\x1b[0m",
         if kind == "create" { "Wrote" } else { "Updated" },
@@ -2459,7 +2497,7 @@ fn format_edit_result(icon: &str, parsed: &serde_json::Value) -> String {
     let path = extract_tool_path(parsed);
     let suffix = if parsed
         .get("replaceAll")
-        .and_then(|value| value.as_bool())
+        .and_then(serde_json::Value::as_bool)
         .unwrap_or(false)
     {
         " (replace all)"
@@ -2487,7 +2525,7 @@ fn format_edit_result(icon: &str, parsed: &serde_json::Value) -> String {
 fn format_glob_result(icon: &str, parsed: &serde_json::Value) -> String {
     let num_files = parsed
         .get("numFiles")
-        .and_then(|value| value.as_u64())
+        .and_then(serde_json::Value::as_u64)
         .unwrap_or(0);
     let filenames = parsed
         .get("filenames")
@@ -2511,11 +2549,11 @@ fn format_glob_result(icon: &str, parsed: &serde_json::Value) -> String {
 fn format_grep_result(icon: &str, parsed: &serde_json::Value) -> String {
     let num_matches = parsed
         .get("numMatches")
-        .and_then(|value| value.as_u64())
+        .and_then(serde_json::Value::as_u64)
         .unwrap_or(0);
     let num_files = parsed
         .get("numFiles")
-        .and_then(|value| value.as_u64())
+        .and_then(serde_json::Value::as_u64)
         .unwrap_or(0);
     let content = parsed
         .get("content")
@@ -2565,10 +2603,9 @@ 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_tools: &mut BTreeMap<u32, (String, String, String)>,
+    pending_tool: &mut Option<(String, String, String)>,
     streaming_tool_input: bool,
 ) -> Result<(), RuntimeError> {
     match block {
@@ -2593,7 +2630,7 @@ fn push_output_block(
             } else {
                 input.to_string()
             };
-            pending_tools.insert(block_index, (id, name, initial_input));
+            *pending_tool = Some((id, name, initial_input));
         }
     }
     Ok(())
@@ -2604,23 +2641,16 @@ fn response_to_events(
     out: &mut (impl Write + ?Sized),
 ) -> Result<Vec<AssistantEvent>, RuntimeError> {
     let mut events = Vec::new();
-    let mut pending_tools = BTreeMap::new();
+    let mut pending_tool = None;
 
-    for (index, block) in response.content.into_iter().enumerate() {
+    for block in response.content {
-        let index =
+        push_output_block(block, out, &mut events, &mut pending_tool, false)?;
-            u32::try_from(index).map_err(|_| RuntimeError::new("response block index overflow"))?;
+        if let Some((id, name, input)) = pending_tool.take() {
-        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 });
         }
     }
 
-    events.push(AssistantEvent::Usage(TokenUsage {
+    events.push(AssistantEvent::Usage(response.usage.token_usage()));
-        input_tokens: response.usage.input_tokens,
-        output_tokens: response.usage.output_tokens,
-        cache_creation_input_tokens: response.usage.cache_creation_input_tokens,
-        cache_read_input_tokens: response.usage.cache_read_input_tokens,
-    }));
     events.push(AssistantEvent::MessageStop);
     Ok(events)
 }
@@ -2836,7 +2866,6 @@ mod tests {
     use api::{MessageResponse, OutputContentBlock, Usage};
     use runtime::{AssistantEvent, ContentBlock, ConversationMessage, MessageRole, PermissionMode};
     use serde_json::json;
-    use std::collections::BTreeMap;
     use std::path::PathBuf;
 
     #[test]
@@ -2916,9 +2945,6 @@ 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");
     }
 
@@ -3386,16 +3412,15 @@ mod tests {
     fn push_output_block_renders_markdown_text() {
         let mut out = Vec::new();
         let mut events = Vec::new();
-        let mut pending_tools = BTreeMap::new();
+        let mut pending_tool = None;
 
         push_output_block(
             OutputContentBlock::Text {
                 text: "# Heading".to_string(),
             },
-            0,
             &mut out,
             &mut events,
-            &mut pending_tools,
+            &mut pending_tool,
             false,
         )
         .expect("text block should render");
@@ -3409,7 +3434,7 @@ 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_tools = BTreeMap::new();
+        let mut pending_tool = None;
 
         push_output_block(
             OutputContentBlock::ToolUse {
@@ -3417,83 +3442,20 @@ mod tests {
                 name: "read_file".to_string(),
                 input: json!({}),
             },
-            1,
             &mut out,
             &mut events,
-            &mut pending_tools,
+            &mut pending_tool,
             true,
         )
         .expect("tool block should accumulate");
 
         assert!(events.is_empty());
         assert_eq!(
-            pending_tools.remove(&1),
+            pending_tool,
             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/rusty-claude-cli/src/render.rs

@@ -286,7 +286,7 @@ impl TerminalRenderer {
     ) {
         match event {
             Event::Start(Tag::Heading { level, .. }) => {
-                self.start_heading(state, level as u8, output)
+                Self::start_heading(state, level as u8, output);
             }
             Event::End(TagEnd::Paragraph) => output.push_str("\n\n"),
             Event::Start(Tag::BlockQuote(..)) => self.start_quote(state, output),
@@ -426,7 +426,7 @@ impl TerminalRenderer {
         }
     }
 
-    fn start_heading(&self, state: &mut RenderState, level: u8, output: &mut String) {
+    fn start_heading(state: &mut RenderState, level: u8, output: &mut String) {
         state.heading_level = Some(level);
         if !output.is_empty() {
             output.push('\n');

rust/crates/telemetry/Cargo.toml

@@ -0,0 +1,13 @@
+[package]
+name = "telemetry"
+version.workspace = true
+edition.workspace = true
+license.workspace = true
+publish.workspace = true
+
+[dependencies]
+serde = { version = "1", features = ["derive"] }
+serde_json = "1"
+
+[lints]
+workspace = true

rust/crates/telemetry/src/lib.rs

@@ -0,0 +1,526 @@
+use std::fmt::{Debug, Formatter};
+use std::fs::{File, OpenOptions};
+use std::io::Write;
+use std::path::{Path, PathBuf};
+use std::sync::atomic::{AtomicU64, Ordering};
+use std::sync::{Arc, Mutex};
+use std::time::{SystemTime, UNIX_EPOCH};
+
+use serde::{Deserialize, Serialize};
+use serde_json::{Map, Value};
+
+pub const DEFAULT_ANTHROPIC_VERSION: &str = "2023-06-01";
+pub const DEFAULT_APP_NAME: &str = "claude-code";
+pub const DEFAULT_RUNTIME: &str = "rust";
+pub const DEFAULT_AGENTIC_BETA: &str = "claude-code-20250219";
+pub const DEFAULT_PROMPT_CACHING_SCOPE_BETA: &str = "prompt-caching-scope-2026-01-05";
+
+#[derive(Debug, Clone, PartialEq, Eq, Serialize, Deserialize)]
+pub struct ClientIdentity {
+    pub app_name: String,
+    pub app_version: String,
+    pub runtime: String,
+}
+
+impl ClientIdentity {
+    #[must_use]
+    pub fn new(app_name: impl Into<String>, app_version: impl Into<String>) -> Self {
+        Self {
+            app_name: app_name.into(),
+            app_version: app_version.into(),
+            runtime: DEFAULT_RUNTIME.to_string(),
+        }
+    }
+
+    #[must_use]
+    pub fn with_runtime(mut self, runtime: impl Into<String>) -> Self {
+        self.runtime = runtime.into();
+        self
+    }
+
+    #[must_use]
+    pub fn user_agent(&self) -> String {
+        format!("{}/{}", self.app_name, self.app_version)
+    }
+}
+
+impl Default for ClientIdentity {
+    fn default() -> Self {
+        Self::new(DEFAULT_APP_NAME, env!("CARGO_PKG_VERSION"))
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub struct AnthropicRequestProfile {
+    pub anthropic_version: String,
+    pub client_identity: ClientIdentity,
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub betas: Vec<String>,
+    #[serde(default, skip_serializing_if = "Map::is_empty")]
+    pub extra_body: Map<String, Value>,
+}
+
+impl AnthropicRequestProfile {
+    #[must_use]
+    pub fn new(client_identity: ClientIdentity) -> Self {
+        Self {
+            anthropic_version: DEFAULT_ANTHROPIC_VERSION.to_string(),
+            client_identity,
+            betas: vec![
+                DEFAULT_AGENTIC_BETA.to_string(),
+                DEFAULT_PROMPT_CACHING_SCOPE_BETA.to_string(),
+            ],
+            extra_body: Map::new(),
+        }
+    }
+
+    #[must_use]
+    pub fn with_beta(mut self, beta: impl Into<String>) -> Self {
+        let beta = beta.into();
+        if !self.betas.contains(&beta) {
+            self.betas.push(beta);
+        }
+        self
+    }
+
+    #[must_use]
+    pub fn with_extra_body(mut self, key: impl Into<String>, value: Value) -> Self {
+        self.extra_body.insert(key.into(), value);
+        self
+    }
+
+    #[must_use]
+    pub fn header_pairs(&self) -> Vec<(String, String)> {
+        let mut headers = vec![
+            (
+                "anthropic-version".to_string(),
+                self.anthropic_version.clone(),
+            ),
+            ("user-agent".to_string(), self.client_identity.user_agent()),
+        ];
+        if !self.betas.is_empty() {
+            headers.push(("anthropic-beta".to_string(), self.betas.join(",")));
+        }
+        headers
+    }
+
+    pub fn render_json_body<T: Serialize>(&self, request: &T) -> Result<Value, serde_json::Error> {
+        let mut body = serde_json::to_value(request)?;
+        let object = body.as_object_mut().ok_or_else(|| {
+            serde_json::Error::io(std::io::Error::new(
+                std::io::ErrorKind::InvalidData,
+                "request body must serialize to a JSON object",
+            ))
+        })?;
+        for (key, value) in &self.extra_body {
+            object.insert(key.clone(), value.clone());
+        }
+        if !self.betas.is_empty() {
+            object.insert(
+                "betas".to_string(),
+                Value::Array(self.betas.iter().cloned().map(Value::String).collect()),
+            );
+        }
+        Ok(body)
+    }
+}
+
+impl Default for AnthropicRequestProfile {
+    fn default() -> Self {
+        Self::new(ClientIdentity::default())
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub struct AnalyticsEvent {
+    pub namespace: String,
+    pub action: String,
+    #[serde(default, skip_serializing_if = "Map::is_empty")]
+    pub properties: Map<String, Value>,
+}
+
+impl AnalyticsEvent {
+    #[must_use]
+    pub fn new(namespace: impl Into<String>, action: impl Into<String>) -> Self {
+        Self {
+            namespace: namespace.into(),
+            action: action.into(),
+            properties: Map::new(),
+        }
+    }
+
+    #[must_use]
+    pub fn with_property(mut self, key: impl Into<String>, value: Value) -> Self {
+        self.properties.insert(key.into(), value);
+        self
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+pub struct SessionTraceRecord {
+    pub session_id: String,
+    pub sequence: u64,
+    pub name: String,
+    pub timestamp_ms: u64,
+    #[serde(default, skip_serializing_if = "Map::is_empty")]
+    pub attributes: Map<String, Value>,
+}
+
+#[derive(Debug, Clone, PartialEq, Serialize, Deserialize)]
+#[serde(tag = "type", rename_all = "snake_case")]
+pub enum TelemetryEvent {
+    HttpRequestStarted {
+        session_id: String,
+        attempt: u32,
+        method: String,
+        path: String,
+        #[serde(default, skip_serializing_if = "Map::is_empty")]
+        attributes: Map<String, Value>,
+    },
+    HttpRequestSucceeded {
+        session_id: String,
+        attempt: u32,
+        method: String,
+        path: String,
+        status: u16,
+        #[serde(default, skip_serializing_if = "Option::is_none")]
+        request_id: Option<String>,
+        #[serde(default, skip_serializing_if = "Map::is_empty")]
+        attributes: Map<String, Value>,
+    },
+    HttpRequestFailed {
+        session_id: String,
+        attempt: u32,
+        method: String,
+        path: String,
+        error: String,
+        retryable: bool,
+        #[serde(default, skip_serializing_if = "Map::is_empty")]
+        attributes: Map<String, Value>,
+    },
+    Analytics(AnalyticsEvent),
+    SessionTrace(SessionTraceRecord),
+}
+
+pub trait TelemetrySink: Send + Sync {
+    fn record(&self, event: TelemetryEvent);
+}
+
+#[derive(Default)]
+pub struct MemoryTelemetrySink {
+    events: Mutex<Vec<TelemetryEvent>>,
+}
+
+impl MemoryTelemetrySink {
+    #[must_use]
+    pub fn events(&self) -> Vec<TelemetryEvent> {
+        self.events
+            .lock()
+            .unwrap_or_else(std::sync::PoisonError::into_inner)
+            .clone()
+    }
+}
+
+impl TelemetrySink for MemoryTelemetrySink {
+    fn record(&self, event: TelemetryEvent) {
+        self.events
+            .lock()
+            .unwrap_or_else(std::sync::PoisonError::into_inner)
+            .push(event);
+    }
+}
+
+pub struct JsonlTelemetrySink {
+    path: PathBuf,
+    file: Mutex<File>,
+}
+
+impl Debug for JsonlTelemetrySink {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("JsonlTelemetrySink")
+            .field("path", &self.path)
+            .finish_non_exhaustive()
+    }
+}
+
+impl JsonlTelemetrySink {
+    pub fn new(path: impl AsRef<Path>) -> Result<Self, std::io::Error> {
+        let path = path.as_ref().to_path_buf();
+        if let Some(parent) = path.parent() {
+            std::fs::create_dir_all(parent)?;
+        }
+        let file = OpenOptions::new().create(true).append(true).open(&path)?;
+        Ok(Self {
+            path,
+            file: Mutex::new(file),
+        })
+    }
+
+    #[must_use]
+    pub fn path(&self) -> &Path {
+        &self.path
+    }
+}
+
+impl TelemetrySink for JsonlTelemetrySink {
+    fn record(&self, event: TelemetryEvent) {
+        let Ok(line) = serde_json::to_string(&event) else {
+            return;
+        };
+        let mut file = self
+            .file
+            .lock()
+            .unwrap_or_else(std::sync::PoisonError::into_inner);
+        let _ = writeln!(file, "{line}");
+        let _ = file.flush();
+    }
+}
+
+#[derive(Clone)]
+pub struct SessionTracer {
+    session_id: String,
+    sequence: Arc<AtomicU64>,
+    sink: Arc<dyn TelemetrySink>,
+}
+
+impl Debug for SessionTracer {
+    fn fmt(&self, f: &mut Formatter<'_>) -> std::fmt::Result {
+        f.debug_struct("SessionTracer")
+            .field("session_id", &self.session_id)
+            .finish_non_exhaustive()
+    }
+}
+
+impl SessionTracer {
+    #[must_use]
+    pub fn new(session_id: impl Into<String>, sink: Arc<dyn TelemetrySink>) -> Self {
+        Self {
+            session_id: session_id.into(),
+            sequence: Arc::new(AtomicU64::new(0)),
+            sink,
+        }
+    }
+
+    #[must_use]
+    pub fn session_id(&self) -> &str {
+        &self.session_id
+    }
+
+    pub fn record(&self, name: impl Into<String>, attributes: Map<String, Value>) {
+        let record = SessionTraceRecord {
+            session_id: self.session_id.clone(),
+            sequence: self.sequence.fetch_add(1, Ordering::Relaxed),
+            name: name.into(),
+            timestamp_ms: current_timestamp_ms(),
+            attributes,
+        };
+        self.sink.record(TelemetryEvent::SessionTrace(record));
+    }
+
+    pub fn record_http_request_started(
+        &self,
+        attempt: u32,
+        method: impl Into<String>,
+        path: impl Into<String>,
+        attributes: Map<String, Value>,
+    ) {
+        let method = method.into();
+        let path = path.into();
+        self.sink.record(TelemetryEvent::HttpRequestStarted {
+            session_id: self.session_id.clone(),
+            attempt,
+            method: method.clone(),
+            path: path.clone(),
+            attributes: attributes.clone(),
+        });
+        self.record(
+            "http_request_started",
+            merge_trace_fields(method, path, attempt, attributes),
+        );
+    }
+
+    pub fn record_http_request_succeeded(
+        &self,
+        attempt: u32,
+        method: impl Into<String>,
+        path: impl Into<String>,
+        status: u16,
+        request_id: Option<String>,
+        attributes: Map<String, Value>,
+    ) {
+        let method = method.into();
+        let path = path.into();
+        self.sink.record(TelemetryEvent::HttpRequestSucceeded {
+            session_id: self.session_id.clone(),
+            attempt,
+            method: method.clone(),
+            path: path.clone(),
+            status,
+            request_id: request_id.clone(),
+            attributes: attributes.clone(),
+        });
+        let mut trace_attributes = merge_trace_fields(method, path, attempt, attributes);
+        trace_attributes.insert("status".to_string(), Value::from(status));
+        if let Some(request_id) = request_id {
+            trace_attributes.insert("request_id".to_string(), Value::String(request_id));
+        }
+        self.record("http_request_succeeded", trace_attributes);
+    }
+
+    pub fn record_http_request_failed(
+        &self,
+        attempt: u32,
+        method: impl Into<String>,
+        path: impl Into<String>,
+        error: impl Into<String>,
+        retryable: bool,
+        attributes: Map<String, Value>,
+    ) {
+        let method = method.into();
+        let path = path.into();
+        let error = error.into();
+        self.sink.record(TelemetryEvent::HttpRequestFailed {
+            session_id: self.session_id.clone(),
+            attempt,
+            method: method.clone(),
+            path: path.clone(),
+            error: error.clone(),
+            retryable,
+            attributes: attributes.clone(),
+        });
+        let mut trace_attributes = merge_trace_fields(method, path, attempt, attributes);
+        trace_attributes.insert("error".to_string(), Value::String(error));
+        trace_attributes.insert("retryable".to_string(), Value::Bool(retryable));
+        self.record("http_request_failed", trace_attributes);
+    }
+
+    pub fn record_analytics(&self, event: AnalyticsEvent) {
+        let mut attributes = event.properties.clone();
+        attributes.insert(
+            "namespace".to_string(),
+            Value::String(event.namespace.clone()),
+        );
+        attributes.insert("action".to_string(), Value::String(event.action.clone()));
+        self.sink.record(TelemetryEvent::Analytics(event));
+        self.record("analytics", attributes);
+    }
+}
+
+fn merge_trace_fields(
+    method: String,
+    path: String,
+    attempt: u32,
+    mut attributes: Map<String, Value>,
+) -> Map<String, Value> {
+    attributes.insert("method".to_string(), Value::String(method));
+    attributes.insert("path".to_string(), Value::String(path));
+    attributes.insert("attempt".to_string(), Value::from(attempt));
+    attributes
+}
+
+fn current_timestamp_ms() -> u64 {
+    SystemTime::now()
+        .duration_since(UNIX_EPOCH)
+        .unwrap_or_default()
+        .as_millis()
+        .try_into()
+        .unwrap_or(u64::MAX)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn request_profile_emits_headers_and_merges_body() {
+        let profile = AnthropicRequestProfile::new(
+            ClientIdentity::new("claude-code", "1.2.3").with_runtime("rust-cli"),
+        )
+        .with_beta("tools-2026-04-01")
+        .with_extra_body("metadata", serde_json::json!({"source": "test"}));
+
+        assert_eq!(
+            profile.header_pairs(),
+            vec![
+                (
+                    "anthropic-version".to_string(),
+                    DEFAULT_ANTHROPIC_VERSION.to_string()
+                ),
+                ("user-agent".to_string(), "claude-code/1.2.3".to_string()),
+                (
+                    "anthropic-beta".to_string(),
+                    "claude-code-20250219,prompt-caching-scope-2026-01-05,tools-2026-04-01"
+                        .to_string(),
+                ),
+            ]
+        );
+
+        let body = profile
+            .render_json_body(&serde_json::json!({"model": "claude-sonnet"}))
+            .expect("body should serialize");
+        assert_eq!(
+            body["metadata"]["source"],
+            Value::String("test".to_string())
+        );
+        assert_eq!(
+            body["betas"],
+            serde_json::json!([
+                "claude-code-20250219",
+                "prompt-caching-scope-2026-01-05",
+                "tools-2026-04-01"
+            ])
+        );
+    }
+
+    #[test]
+    fn session_tracer_records_structured_events_and_trace_sequence() {
+        let sink = Arc::new(MemoryTelemetrySink::default());
+        let tracer = SessionTracer::new("session-123", sink.clone());
+
+        tracer.record_http_request_started(1, "POST", "/v1/messages", Map::new());
+        tracer.record_analytics(
+            AnalyticsEvent::new("cli", "prompt_sent")
+                .with_property("model", Value::String("claude-opus".to_string())),
+        );
+
+        let events = sink.events();
+        assert!(matches!(
+            &events[0],
+            TelemetryEvent::HttpRequestStarted {
+                session_id,
+                attempt: 1,
+                method,
+                path,
+                ..
+            } if session_id == "session-123" && method == "POST" && path == "/v1/messages"
+        ));
+        assert!(matches!(
+            &events[1],
+            TelemetryEvent::SessionTrace(SessionTraceRecord { sequence: 0, name, .. })
+            if name == "http_request_started"
+        ));
+        assert!(matches!(&events[2], TelemetryEvent::Analytics(_)));
+        assert!(matches!(
+            &events[3],
+            TelemetryEvent::SessionTrace(SessionTraceRecord { sequence: 1, name, .. })
+            if name == "analytics"
+        ));
+    }
+
+    #[test]
+    fn jsonl_sink_persists_events() {
+        let path =
+            std::env::temp_dir().join(format!("telemetry-jsonl-{}.log", current_timestamp_ms()));
+        let sink = JsonlTelemetrySink::new(&path).expect("sink should create file");
+
+        sink.record(TelemetryEvent::Analytics(
+            AnalyticsEvent::new("cli", "turn_completed").with_property("ok", Value::Bool(true)),
+        ));
+
+        let contents = std::fs::read_to_string(&path).expect("telemetry log should be readable");
+        assert!(contents.contains("\"type\":\"analytics\""));
+        assert!(contents.contains("\"action\":\"turn_completed\""));
+
+        let _ = std::fs::remove_file(path);
+    }
+}

rust/crates/tools/src/lib.rs

@@ -4,16 +4,16 @@ use std::process::Command;
 use std::time::{Duration, Instant};
 
 use api::{
-    max_tokens_for_model, resolve_model_alias, ContentBlockDelta, InputContentBlock, InputMessage,
+    read_base_url, AnthropicClient, ContentBlockDelta, InputContentBlock, InputMessage,
-    MessageRequest, MessageResponse, OutputContentBlock, ProviderClient,
+    MessageRequest, MessageResponse, OutputContentBlock, StreamEvent as ApiStreamEvent, ToolChoice,
-    StreamEvent as ApiStreamEvent, ToolChoice, ToolDefinition, ToolResultContentBlock,
+    ToolDefinition, ToolResultContentBlock,
 };
 use reqwest::blocking::Client;
 use runtime::{
     edit_file, execute_bash, glob_search, grep_search, load_system_prompt, read_file, write_file,
     ApiClient, ApiRequest, AssistantEvent, BashCommandInput, ContentBlock, ConversationMessage,
     ConversationRuntime, GrepSearchInput, MessageRole, PermissionMode, PermissionPolicy,
-    RuntimeError, Session, TokenUsage, ToolError, ToolExecutor,
+    RuntimeError, Session, ToolError, ToolExecutor,
 };
 use serde::{Deserialize, Serialize};
 use serde_json::{json, Value};
@@ -1459,14 +1459,14 @@ fn run_agent_job(job: &AgentJob) -> Result<(), String> {
 
 fn build_agent_runtime(
     job: &AgentJob,
-) -> Result<ConversationRuntime<ProviderRuntimeClient, SubagentToolExecutor>, String> {
+) -> Result<ConversationRuntime<AnthropicRuntimeClient, 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 = ProviderRuntimeClient::new(model, allowed_tools.clone())?;
+    let api_client = AnthropicRuntimeClient::new(model, allowed_tools.clone())?;
     let tool_executor = SubagentToolExecutor::new(allowed_tools);
     Ok(ConversationRuntime::new(
         Session::new(),
@@ -1635,17 +1635,18 @@ fn format_agent_terminal_output(status: &str, result: Option<&str>, error: Optio
     sections.join("")
 }
 
-struct ProviderRuntimeClient {
+struct AnthropicRuntimeClient {
     runtime: tokio::runtime::Runtime,
-    client: ProviderClient,
+    client: AnthropicClient,
     model: String,
     allowed_tools: BTreeSet<String>,
 }
 
-impl ProviderRuntimeClient {
+impl AnthropicRuntimeClient {
     fn new(model: String, allowed_tools: BTreeSet<String>) -> Result<Self, String> {
-        let model = resolve_model_alias(&model).to_string();
+        let client = AnthropicClient::from_env()
-        let client = ProviderClient::from_model(&model).map_err(|error| error.to_string())?;
+            .map_err(|error| error.to_string())?
+            .with_base_url(read_base_url());
         Ok(Self {
             runtime: tokio::runtime::Runtime::new().map_err(|error| error.to_string())?,
             client,
@@ -1655,7 +1656,7 @@ impl ProviderRuntimeClient {
     }
 }
 
-impl ApiClient for ProviderRuntimeClient {
+impl ApiClient for AnthropicRuntimeClient {
     fn stream(&mut self, request: ApiRequest) -> Result<Vec<AssistantEvent>, RuntimeError> {
         let tools = tool_specs_for_allowed_tools(Some(&self.allowed_tools))
             .into_iter()
@@ -1667,7 +1668,7 @@ impl ApiClient for ProviderRuntimeClient {
             .collect::<Vec<_>>();
         let message_request = MessageRequest {
             model: self.model.clone(),
-            max_tokens: max_tokens_for_model(&self.model),
+            max_tokens: 32_000,
             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),
@@ -1682,7 +1683,7 @@ impl ApiClient for ProviderRuntimeClient {
                 .await
                 .map_err(|error| RuntimeError::new(error.to_string()))?;
             let mut events = Vec::new();
-            let mut pending_tools: BTreeMap<u32, (String, String, String)> = BTreeMap::new();
+            let mut pending_tool: Option<(String, String, String)> = None;
             let mut saw_stop = false;
 
             while let Some(event) = stream
@@ -1693,15 +1694,14 @@ impl ApiClient for ProviderRuntimeClient {
                 match event {
                     ApiStreamEvent::MessageStart(start) => {
                         for block in start.message.content {
-                            push_output_block(block, 0, &mut events, &mut pending_tools, true);
+                            push_output_block(block, &mut events, &mut pending_tool, true);
                         }
                     }
                     ApiStreamEvent::ContentBlockStart(start) => {
                         push_output_block(
                             start.content_block,
-                            start.index,
                             &mut events,
-                            &mut pending_tools,
+                            &mut pending_tool,
                             true,
                         );
                     }
@@ -1712,23 +1712,18 @@ impl ApiClient for ProviderRuntimeClient {
                             }
                         }
                         ContentBlockDelta::InputJsonDelta { partial_json } => {
-                            if let Some((_, _, input)) = pending_tools.get_mut(&delta.index) {
+                            if let Some((_, _, input)) = &mut pending_tool {
                                 input.push_str(&partial_json);
                             }
                         }
                     },
-                    ApiStreamEvent::ContentBlockStop(stop) => {
+                    ApiStreamEvent::ContentBlockStop(_) => {
-                        if let Some((id, name, input)) = pending_tools.remove(&stop.index) {
+                        if let Some((id, name, input)) = pending_tool.take() {
                             events.push(AssistantEvent::ToolUse { id, name, input });
                         }
                     }
                     ApiStreamEvent::MessageDelta(delta) => {
-                        events.push(AssistantEvent::Usage(TokenUsage {
+                        events.push(AssistantEvent::Usage(delta.usage.token_usage()));
-                            input_tokens: delta.usage.input_tokens,
-                            output_tokens: delta.usage.output_tokens,
-                            cache_creation_input_tokens: 0,
-                            cache_read_input_tokens: 0,
-                        }));
                     }
                     ApiStreamEvent::MessageStop(_) => {
                         saw_stop = true;
@@ -1839,9 +1834,8 @@ fn convert_messages(messages: &[ConversationMessage]) -> Vec<InputMessage> {
 
 fn push_output_block(
     block: OutputContentBlock,
-    block_index: u32,
     events: &mut Vec<AssistantEvent>,
-    pending_tools: &mut BTreeMap<u32, (String, String, String)>,
+    pending_tool: &mut Option<(String, String, String)>,
     streaming_tool_input: bool,
 ) {
     match block {
@@ -1859,29 +1853,23 @@ fn push_output_block(
             } else {
                 input.to_string()
             };
-            pending_tools.insert(block_index, (id, name, initial_input));
+            *pending_tool = Some((id, name, initial_input));
         }
     }
 }
 
 fn response_to_events(response: MessageResponse) -> Vec<AssistantEvent> {
     let mut events = Vec::new();
-    let mut pending_tools = BTreeMap::new();
+    let mut pending_tool = None;
 
-    for (index, block) in response.content.into_iter().enumerate() {
+    for block in response.content {
-        let index = u32::try_from(index).expect("response block index overflow");
+        push_output_block(block, &mut events, &mut pending_tool, false);
-        push_output_block(block, index, &mut events, &mut pending_tools, false);
+        if let Some((id, name, input)) = pending_tool.take() {
-        if let Some((id, name, input)) = pending_tools.remove(&index) {
             events.push(AssistantEvent::ToolUse { id, name, input });
         }
     }
 
-    events.push(AssistantEvent::Usage(TokenUsage {
+    events.push(AssistantEvent::Usage(response.usage.token_usage()));
-        input_tokens: response.usage.input_tokens,
-        output_tokens: response.usage.output_tokens,
-        cache_creation_input_tokens: response.usage.cache_creation_input_tokens,
-        cache_read_input_tokens: response.usage.cache_read_input_tokens,
-    }));
     events.push(AssistantEvent::MessageStop);
     events
 }
@@ -2895,7 +2883,6 @@ 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};
@@ -2908,9 +2895,8 @@ 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,
-        push_output_block, AgentInput, AgentJob, SubagentToolExecutor,
+        AgentInput, AgentJob, SubagentToolExecutor,
     };
-    use api::OutputContentBlock;
     use runtime::{ApiRequest, AssistantEvent, ConversationRuntime, RuntimeError, Session};
     use serde_json::json;
 
@@ -3125,63 +3111,6 @@ 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()