fix: add agent_server_logs module import to lib.rs

server/CLAUDE.md

@@ -1,12 +1,27 @@
 # Server Testing
 
+## Test placement
+
+Place all new tests under `server/packages/**/tests/` (or a package-specific `tests/` folder). Avoid inline tests inside source files unless there is no viable alternative.
+
+## Test locations (overview)
+
+- Sandbox-agent integration tests live under `server/packages/sandbox-agent/tests/`:
+  - Agent flow coverage in `agent-flows/`
+  - Agent management coverage in `agent-management/`
+  - Shared server manager coverage in `server-manager/`
+  - HTTP/SSE and snapshot coverage in `http/` (snapshots in `http/snapshots/`)
+  - UI coverage in `ui/`
+  - Shared helpers in `common/`
+- Extracted agent schema roundtrip tests live under `server/packages/extracted-agent-schemas/tests/`
+
 ## Snapshot tests
 
-The HTTP/SSE snapshot suite lives in:
-- `server/packages/sandbox-agent/tests/http_sse_snapshots.rs`
+The HTTP/SSE snapshot suite entrypoint lives in:
+- `server/packages/sandbox-agent/tests/http_sse_snapshots.rs` (includes `tests/http/http_sse_snapshots.rs`)
 
 Snapshots are written to:
-- `server/packages/sandbox-agent/tests/snapshots/`
+- `server/packages/sandbox-agent/tests/http/snapshots/`
 
 ## Agent selection
 
@@ -47,9 +62,20 @@ Health checks run in a blocking thread to avoid Tokio runtime drop errors inside
 ## Snapshot stability
 
 To keep snapshots deterministic:
+- Use the mock agent as the **master** event sequence; all other agents must match its behavior 1:1.
+- Snapshots should compare a **canonical event skeleton** (event order matters) with strict ordering across:
+  - `item.started` → `item.delta` → `item.completed`
+  - presence/absence of `session.ended`
+  - permission/question request and resolution flows
+- Scrub non-deterministic fields from snapshots:
+  - IDs, timestamps, native IDs
+  - text content, tool inputs/outputs, provider-specific metadata
+  - `source` and `synthetic` flags (these are implementation details)
+- The sandbox-agent is responsible for emitting **synthetic events** so that real agents match the mock sequence exactly.
 - Event streams are truncated after the first assistant or error event.
 - Permission flow snapshots are truncated after the permission request (or first assistant) event.
 - Unknown events are preserved as `kind: unknown` (raw payload in universal schema).
+- Prefer snapshot-based event skeleton assertions over manual event-order assertions in tests.
 
 ## Typical commands
 

server/packages/extracted-agent-schemas/src/lib.rs

@@ -25,87 +25,3 @@ pub mod amp {
     //! AMP Code SDK types.
     include!(concat!(env!("OUT_DIR"), "/amp.rs"));
 }
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn test_claude_bash_input() {
-        let input = claude::BashInput {
-            command: "ls -la".to_string(),
-            timeout: Some(5000.0),
-            working_directory: None,
-        };
-
-        let json = serde_json::to_string(&input).unwrap();
-        assert!(json.contains("ls -la"));
-
-        let parsed: claude::BashInput = serde_json::from_str(&json).unwrap();
-        assert_eq!(parsed.command, "ls -la");
-    }
-
-    #[test]
-    fn test_codex_server_notification() {
-        // Test ItemCompletedNotification with AgentMessage
-        let notification = codex::ServerNotification::ItemCompleted(
-            codex::ItemCompletedNotification {
-                item: codex::ThreadItem::AgentMessage {
-                    id: "msg-123".to_string(),
-                    text: "Hello from Codex".to_string(),
-                },
-                thread_id: "thread-123".to_string(),
-                turn_id: "turn-456".to_string(),
-            }
-        );
-
-        let json = serde_json::to_string(&notification).unwrap();
-        assert!(json.contains("item/completed"));
-        assert!(json.contains("Hello from Codex"));
-        assert!(json.contains("agentMessage"));
-    }
-
-    #[test]
-    fn test_codex_thread_item_variants() {
-        // Test UserMessage variant
-        let user_msg = codex::ThreadItem::UserMessage {
-            content: vec![codex::UserInput::Text {
-                text: "Hello".to_string(),
-                text_elements: vec![],
-            }],
-            id: "user-1".to_string(),
-        };
-        let json = serde_json::to_string(&user_msg).unwrap();
-        assert!(json.contains("userMessage"));
-        assert!(json.contains("Hello"));
-
-        // Test CommandExecution variant
-        let cmd = codex::ThreadItem::CommandExecution {
-            aggregated_output: Some("output".to_string()),
-            command: "ls -la".to_string(),
-            command_actions: vec![],
-            cwd: "/tmp".to_string(),
-            duration_ms: Some(100),
-            exit_code: Some(0),
-            id: "cmd-1".to_string(),
-            process_id: None,
-            status: codex::CommandExecutionStatus::Completed,
-        };
-        let json = serde_json::to_string(&cmd).unwrap();
-        assert!(json.contains("commandExecution"));
-        assert!(json.contains("ls -la"));
-    }
-
-    #[test]
-    fn test_amp_message() {
-        let msg = amp::Message {
-            role: amp::MessageRole::User,
-            content: "Hello".to_string(),
-            tool_calls: vec![],
-        };
-
-        let json = serde_json::to_string(&msg).unwrap();
-        assert!(json.contains("user"));
-        assert!(json.contains("Hello"));
-    }
-}

server/packages/extracted-agent-schemas/tests/schema_roundtrip.rs

@@ -0,0 +1,77 @@
+use sandbox_agent_extracted_agent_schemas::{amp, claude, codex};
+
+#[test]
+fn test_claude_bash_input() {
+    let input = claude::BashInput {
+        command: "ls -la".to_string(),
+        timeout: Some(5000.0),
+        working_directory: None,
+    };
+
+    let json = serde_json::to_string(&input).unwrap();
+    assert!(json.contains("ls -la"));
+
+    let parsed: claude::BashInput = serde_json::from_str(&json).unwrap();
+    assert_eq!(parsed.command, "ls -la");
+}
+
+#[test]
+fn test_codex_server_notification() {
+    let notification = codex::ServerNotification::ItemCompleted(
+        codex::ItemCompletedNotification {
+            item: codex::ThreadItem::AgentMessage {
+                id: "msg-123".to_string(),
+                text: "Hello from Codex".to_string(),
+            },
+            thread_id: "thread-123".to_string(),
+            turn_id: "turn-456".to_string(),
+        },
+    );
+
+    let json = serde_json::to_string(&notification).unwrap();
+    assert!(json.contains("item/completed"));
+    assert!(json.contains("Hello from Codex"));
+    assert!(json.contains("agentMessage"));
+}
+
+#[test]
+fn test_codex_thread_item_variants() {
+    let user_msg = codex::ThreadItem::UserMessage {
+        content: vec![codex::UserInput::Text {
+            text: "Hello".to_string(),
+            text_elements: vec![],
+        }],
+        id: "user-1".to_string(),
+    };
+    let json = serde_json::to_string(&user_msg).unwrap();
+    assert!(json.contains("userMessage"));
+    assert!(json.contains("Hello"));
+
+    let cmd = codex::ThreadItem::CommandExecution {
+        aggregated_output: Some("output".to_string()),
+        command: "ls -la".to_string(),
+        command_actions: vec![],
+        cwd: "/tmp".to_string(),
+        duration_ms: Some(100),
+        exit_code: Some(0),
+        id: "cmd-1".to_string(),
+        process_id: None,
+        status: codex::CommandExecutionStatus::Completed,
+    };
+    let json = serde_json::to_string(&cmd).unwrap();
+    assert!(json.contains("commandExecution"));
+    assert!(json.contains("ls -la"));
+}
+
+#[test]
+fn test_amp_message() {
+    let msg = amp::Message {
+        role: amp::MessageRole::User,
+        content: "Hello".to_string(),
+        tool_calls: vec![],
+    };
+
+    let json = serde_json::to_string(&msg).unwrap();
+    assert!(json.contains("user"));
+    assert!(json.contains("Hello"));
+}

server/packages/sandbox-agent/Cargo.toml

@@ -32,9 +32,13 @@ tracing.workspace = true
 tracing-logfmt.workspace = true
 tracing-subscriber.workspace = true
 include_dir.workspace = true
+tempfile = { workspace = true, optional = true }
 
 [dev-dependencies]
 http-body-util.workspace = true
 insta.workspace = true
-tempfile.workspace = true
 tower.workspace = true
+tempfile.workspace = true
+
+[features]
+test-utils = ["tempfile"]

server/packages/sandbox-agent/src/lib.rs

@@ -1,6 +1,7 @@
 //! Sandbox agent core utilities.
 
 pub mod credentials;
+mod agent_server_logs;
 pub mod router;
 pub mod telemetry;
 pub mod ui;

server/packages/sandbox-agent/src/main.rs

@@ -1,6 +1,7 @@
 use std::collections::HashMap;
 use std::io::Write;
 use std::path::PathBuf;
+use std::sync::Arc;
 
 use clap::{Args, Parser, Subcommand};
 use reqwest::blocking::Client as HttpClient;
@@ -16,7 +17,7 @@ use sandbox_agent::router::{
 };
 use sandbox_agent::telemetry;
 use sandbox_agent::router::{AgentListResponse, AgentModesResponse, CreateSessionResponse, EventsResponse};
-use sandbox_agent::router::build_router;
+use sandbox_agent::router::{build_router_with_state, shutdown_servers};
 use sandbox_agent::ui;
 use serde::Serialize;
 use serde_json::Value;
@@ -352,8 +353,8 @@ fn run_server(cli: &Cli, server: &ServerArgs) -> Result<(), CliError> {
 
     let agent_manager =
         AgentManager::new(default_install_dir()).map_err(|err| CliError::Server(err.to_string()))?;
-    let state = AppState::new(auth, agent_manager);
-    let mut router = build_router(state);
+    let state = Arc::new(AppState::new(auth, agent_manager));
+    let (mut router, state) = build_router_with_state(state);
 
     if let Some(cors) = build_cors_layer(server)? {
         router = router.layer(cors);
@@ -384,7 +385,12 @@ fn run_server(cli: &Cli, server: &ServerArgs) -> Result<(), CliError> {
         } else {
             tracing::info!("inspector ui not embedded; set SANDBOX_AGENT_SKIP_INSPECTOR=1 to skip embedding during builds");
         }
+        let shutdown_state = state.clone();
         axum::serve(listener, router)
+            .with_graceful_shutdown(async move {
+                let _ = tokio::signal::ctrl_c().await;
+                shutdown_servers(&shutdown_state).await;
+            })
             .await
             .map_err(|err| CliError::Server(err.to_string()))
     })

server/packages/sandbox-agent/tests/agent-flows/agent_basic_reply.rs

@@ -1,3 +1,4 @@
+#[path = "../common/mod.rs"]
 mod common;
 
 use common::*;
@@ -29,8 +30,6 @@ async fn agent_basic_reply() {
             "no events collected for {}",
             config.agent.as_str()
         );
-        expect_basic_sequence(&events);
-
         let caps = capabilities
             .get(config.agent.as_str())
             .expect("capabilities missing");

server/packages/sandbox-agent/tests/agent-flows/agent_permission_flow.rs

@@ -1,3 +1,4 @@
+#[path = "../common/mod.rs"]
 mod common;
 
 use common::*;

server/packages/sandbox-agent/tests/agent-flows/agent_question_flow.rs

@@ -1,3 +1,4 @@
+#[path = "../common/mod.rs"]
 mod common;
 
 use common::*;

server/packages/sandbox-agent/tests/agent-flows/agent_termination.rs

@@ -1,3 +1,4 @@
+#[path = "../common/mod.rs"]
 mod common;
 
 use common::*;

server/packages/sandbox-agent/tests/agent-flows/agent_tool_flow.rs

@@ -1,3 +1,4 @@
+#[path = "../common/mod.rs"]
 mod common;
 
 use common::*;

server/packages/sandbox-agent/tests/agent-flows/mod.rs

@@ -0,0 +1,6 @@
+mod agent_basic_reply;
+mod agent_multi_turn;
+mod agent_permission_flow;
+mod agent_question_flow;
+mod agent_termination;
+mod agent_tool_flow;

server/packages/sandbox-agent/tests/agent-management/mod.rs

@@ -0,0 +1 @@
+mod agents;

server/packages/sandbox-agent/tests/agent_flows.rs

@@ -0,0 +1,2 @@
+#[path = "agent-flows/mod.rs"]
+mod agent_flows;

server/packages/sandbox-agent/tests/agent_management.rs

@@ -0,0 +1,2 @@
+#[path = "agent-management/mod.rs"]
+mod agent_management;

server/packages/sandbox-agent/tests/common/mod.rs

@@ -272,38 +272,6 @@ pub fn find_assistant_message_item(events: &[Value]) -> Option<String> {
     })
 }
 
-pub fn event_sequence(event: &Value) -> Option<u64> {
-    event.get("sequence").and_then(Value::as_u64)
-}
-
-pub fn find_item_event_seq(events: &[Value], event_type: &str, item_id: &str) -> Option<u64> {
-    events.iter().find_map(|event| {
-        if event.get("type").and_then(Value::as_str) != Some(event_type) {
-            return None;
-        }
-        match event_type {
-            "item.delta" => {
-                let data = event.get("data")?;
-                let id = data.get("item_id")?.as_str()?;
-                if id == item_id {
-                    event_sequence(event)
-                } else {
-                    None
-                }
-            }
-            _ => {
-                let item = event.get("data")?.get("item")?;
-                let id = item.get("item_id")?.as_str()?;
-                if id == item_id {
-                    event_sequence(event)
-                } else {
-                    None
-                }
-            }
-        }
-    })
-}
-
 pub fn find_permission_id(events: &[Value]) -> Option<String> {
     events.iter().find_map(|event| {
         if event.get("type").and_then(Value::as_str) != Some("permission.requested") {
@@ -372,17 +340,3 @@ pub fn has_tool_result(events: &[Value]) -> bool {
         item.get("kind").and_then(Value::as_str) == Some("tool_result")
     })
 }
-
-pub fn expect_basic_sequence(events: &[Value]) {
-    assert!(has_event_type(events, "session.started"), "session.started missing");
-    let item_id = find_assistant_message_item(events).expect("assistant message missing");
-    let started_seq = find_item_event_seq(events, "item.started", &item_id)
-        .expect("item.started missing");
-    // Intentionally require deltas here to validate our synthetic delta behavior.
-    let delta_seq = find_item_event_seq(events, "item.delta", &item_id)
-        .expect("item.delta missing");
-    let completed_seq = find_item_event_seq(events, "item.completed", &item_id)
-        .expect("item.completed missing");
-    assert!(started_seq < delta_seq, "item.started must precede delta");
-    assert!(delta_seq < completed_seq, "delta must precede completion");
-}

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__api_endpoints_snapshots@agent_install_mock.snap

@@ -0,0 +1,6 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1016
+expression: snapshot_status(status)
+---
+status: 204

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__api_endpoints_snapshots@agent_modes_mock.snap

@@ -0,0 +1,12 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1034
+expression: normalize_agent_modes(&modes)
+---
+modes:
+  - description: true
+    id: build
+    name: Build
+  - description: true
+    id: plan
+    name: Plan

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__api_endpoints_snapshots@create_session_mock.snap

@@ -0,0 +1,7 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1053
+expression: normalize_create_session(&created)
+---
+healthy: true
+nativeSessionId: "<redacted>"

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__api_endpoints_snapshots@send_message_mock.snap

@@ -0,0 +1,6 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1078
+expression: snapshot_status(status)
+---
+status: 204

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__approval_flow_snapshots@permission_events_claude.snap

@@ -0,0 +1,17 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1119
+expression: normalize_events(&permission_events)
+---
+- metadata: true
+  seq: 1
+  session: started
+  source: daemon
+  synthetic: true
+  type: session.started
+- metadata: true
+  seq: 2
+  session: started
+  source: agent
+  synthetic: false
+  type: session.started

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__approval_flow_snapshots@permission_events_mock.snap

@@ -1,19 +1,15 @@
 ---
-source: server/packages/sandbox-agent/tests/http_sse_snapshots.rs
-assertion_line: 1025
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1112
 expression: normalize_events(&permission_events)
 ---
 - metadata: true
   seq: 1
   session: started
-  source: daemon
-  synthetic: true
   type: session.started
 - metadata: true
   seq: 2
   session: started
-  source: agent
-  synthetic: false
   type: session.started
 - item:
     content_types:
@@ -22,16 +18,12 @@ expression: normalize_events(&permission_events)
     role: assistant
     status: in_progress
   seq: 3
-  source: daemon
-  synthetic: true
   type: item.started
 - delta:
     delta: "<redacted>"
     item_id: "<redacted>"
     native_item_id: "<redacted>"
   seq: 4
-  source: daemon
-  synthetic: true
   type: item.delta
 - item:
     content_types:
@@ -40,6 +32,4 @@ expression: normalize_events(&permission_events)
     role: assistant
     status: completed
   seq: 5
-  source: agent
-  synthetic: false
   type: item.completed

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__approval_flow_snapshots@permission_reply_missing_mock.snap

@@ -0,0 +1,11 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1152
+expression: "json!({ \"status\": status.as_u16(), \"payload\": payload, })"
+---
+payload:
+  detail: "invalid request: unknown permission id: missing-permission"
+  status: 400
+  title: Invalid Request
+  type: "urn:sandbox-agent:error:invalid_request"
+status: 400

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__approval_flow_snapshots@question_reject_events_mock.snap

@@ -0,0 +1,35 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1236
+expression: normalize_events(&reject_events)
+---
+- metadata: true
+  seq: 1
+  session: started
+  type: session.started
+- metadata: true
+  seq: 2
+  session: started
+  type: session.started
+- item:
+    content_types:
+      - text
+    kind: message
+    role: assistant
+    status: in_progress
+  seq: 3
+  type: item.started
+- delta:
+    delta: "<redacted>"
+    item_id: "<redacted>"
+    native_item_id: "<redacted>"
+  seq: 4
+  type: item.delta
+- item:
+    content_types:
+      - text
+    kind: message
+    role: assistant
+    status: completed
+  seq: 5
+  type: item.completed

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__approval_flow_snapshots@question_reject_missing_mock.snap

@@ -0,0 +1,11 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1276
+expression: "json!({ \"status\": status.as_u16(), \"payload\": payload, })"
+---
+payload:
+  detail: "invalid request: unknown question id: missing-question"
+  status: 400
+  title: Invalid Request
+  type: "urn:sandbox-agent:error:invalid_request"
+status: 400

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__approval_flow_snapshots@question_reply_events_mock.snap

@@ -0,0 +1,35 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1174
+expression: normalize_events(&question_events)
+---
+- metadata: true
+  seq: 1
+  session: started
+  type: session.started
+- metadata: true
+  seq: 2
+  session: started
+  type: session.started
+- item:
+    content_types:
+      - text
+    kind: message
+    role: assistant
+    status: in_progress
+  seq: 3
+  type: item.started
+- delta:
+    delta: "<redacted>"
+    item_id: "<redacted>"
+    native_item_id: "<redacted>"
+  seq: 4
+  type: item.delta
+- item:
+    content_types:
+      - text
+    kind: message
+    role: assistant
+    status: completed
+  seq: 5
+  type: item.completed

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__approval_flow_snapshots@question_reply_missing_mock.snap

@@ -0,0 +1,11 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1214
+expression: "json!({ \"status\": status.as_u16(), \"payload\": payload, })"
+---
+payload:
+  detail: "invalid request: unknown question id: missing-question"
+  status: 400
+  title: Invalid Request
+  type: "urn:sandbox-agent:error:invalid_request"
+status: 400

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__run_concurrency_snapshot@concurrency_events_claude.snap

@@ -1,6 +1,6 @@
 ---
-source: server/packages/sandbox-agent/tests/http_sse_snapshots.rs
-assertion_line: 1259
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1351
 expression: snapshot
 ---
 session_a:
@@ -23,16 +23,16 @@ session_a:
       role: assistant
       status: in_progress
     seq: 3
-    source: daemon
-    synthetic: true
+    source: agent
+    synthetic: false
     type: item.started
   - delta:
       delta: "<redacted>"
       item_id: "<redacted>"
       native_item_id: "<redacted>"
     seq: 4
-    source: daemon
-    synthetic: true
+    source: agent
+    synthetic: false
     type: item.delta
   - item:
       content_types:
@@ -64,16 +64,16 @@ session_b:
       role: assistant
       status: in_progress
     seq: 3
-    source: daemon
-    synthetic: true
+    source: agent
+    synthetic: false
     type: item.started
   - delta:
       delta: "<redacted>"
       item_id: "<redacted>"
       native_item_id: "<redacted>"
     seq: 4
-    source: daemon
-    synthetic: true
+    source: agent
+    synthetic: false
     type: item.delta
   - item:
       content_types:

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__run_concurrency_snapshot@concurrency_events_mock.snap

@@ -0,0 +1,67 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 1344
+expression: snapshot
+---
+session_a:
+  - metadata: true
+    seq: 1
+    session: started
+    type: session.started
+  - metadata: true
+    seq: 2
+    session: started
+    type: session.started
+  - item:
+      content_types:
+        - text
+      kind: message
+      role: assistant
+      status: in_progress
+    seq: 3
+    type: item.started
+  - delta:
+      delta: "<redacted>"
+      item_id: "<redacted>"
+      native_item_id: "<redacted>"
+    seq: 4
+    type: item.delta
+  - item:
+      content_types:
+        - text
+      kind: message
+      role: assistant
+      status: completed
+    seq: 5
+    type: item.completed
+session_b:
+  - metadata: true
+    seq: 1
+    session: started
+    type: session.started
+  - metadata: true
+    seq: 2
+    session: started
+    type: session.started
+  - item:
+      content_types:
+        - text
+      kind: message
+      role: assistant
+      status: in_progress
+    seq: 3
+    type: item.started
+  - delta:
+      delta: "<redacted>"
+      item_id: "<redacted>"
+      native_item_id: "<redacted>"
+    seq: 4
+    type: item.delta
+  - item:
+      content_types:
+        - text
+      kind: message
+      role: assistant
+      status: completed
+    seq: 5
+    type: item.completed

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__run_http_events_snapshot@http_events_claude.snap

@@ -1,6 +1,6 @@
 ---
-source: server/packages/sandbox-agent/tests/http_sse_snapshots.rs
-assertion_line: 742
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 811
 expression: normalized
 ---
 - metadata: true
@@ -22,16 +22,16 @@ expression: normalized
     role: assistant
     status: in_progress
   seq: 3
-  source: daemon
-  synthetic: true
+  source: agent
+  synthetic: false
   type: item.started
 - delta:
     delta: "<redacted>"
     item_id: "<redacted>"
     native_item_id: "<redacted>"
   seq: 4
-  source: daemon
-  synthetic: true
+  source: agent
+  synthetic: false
   type: item.delta
 - item:
     content_types:

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__run_http_events_snapshot@http_events_mock.snap

@@ -0,0 +1,35 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 804
+expression: normalized
+---
+- metadata: true
+  seq: 1
+  session: started
+  type: session.started
+- metadata: true
+  seq: 2
+  session: started
+  type: session.started
+- item:
+    content_types:
+      - text
+    kind: message
+    role: assistant
+    status: in_progress
+  seq: 3
+  type: item.started
+- delta:
+    delta: "<redacted>"
+    item_id: "<redacted>"
+    native_item_id: "<redacted>"
+  seq: 4
+  type: item.delta
+- item:
+    content_types:
+      - text
+    kind: message
+    role: assistant
+    status: completed
+  seq: 5
+  type: item.completed

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__run_sse_events_snapshot@sse_events_claude.snap

@@ -1,6 +1,6 @@
 ---
-source: server/packages/sandbox-agent/tests/http_sse_snapshots.rs
-assertion_line: 775
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 848
 expression: normalized
 ---
 - metadata: true
@@ -22,16 +22,16 @@ expression: normalized
     role: assistant
     status: in_progress
   seq: 3
-  source: daemon
-  synthetic: true
+  source: agent
+  synthetic: false
   type: item.started
 - delta:
     delta: "<redacted>"
     item_id: "<redacted>"
     native_item_id: "<redacted>"
   seq: 4
-  source: daemon
-  synthetic: true
+  source: agent
+  synthetic: false
   type: item.delta
 - item:
     content_types:

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__run_sse_events_snapshot@sse_events_codex.snap

@@ -1,5 +1,6 @@
 ---
-source: server/packages/sandbox-agent/tests/http_sse_snapshots.rs
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 848
 expression: normalized
 ---
 - metadata: true
@@ -62,30 +63,11 @@ expression: normalized
   synthetic: false
   type: item.started
 - item:
-    content_types:
-      - status
-    kind: status
-    role: system
+    content_types: []
+    kind: message
+    role: assistant
     status: completed
   seq: 8
   source: agent
   synthetic: false
   type: item.completed
-- delta:
-    delta: "<redacted>"
-    item_id: "<redacted>"
-    native_item_id: "<redacted>"
-  seq: 9
-  source: agent
-  synthetic: false
-  type: item.delta
-- item:
-    content_types:
-      - reasoning
-    kind: message
-    role: assistant
-    status: completed
-  seq: 10
-  source: agent
-  synthetic: false
-  type: item.completed

server/packages/sandbox-agent/tests/http/snapshots/http_sse_snapshots__run_sse_events_snapshot@sse_events_mock.snap

@@ -0,0 +1,35 @@
+---
+source: server/packages/sandbox-agent/tests/http/http_sse_snapshots.rs
+assertion_line: 841
+expression: normalized
+---
+- metadata: true
+  seq: 1
+  session: started
+  type: session.started
+- metadata: true
+  seq: 2
+  session: started
+  type: session.started
+- item:
+    content_types:
+      - text
+    kind: message
+    role: assistant
+    status: in_progress
+  seq: 3
+  type: item.started
+- delta:
+    delta: "<redacted>"
+    item_id: "<redacted>"
+    native_item_id: "<redacted>"
+  seq: 4
+  type: item.delta
+- item:
+    content_types:
+      - text
+    kind: message
+    role: assistant
+    status: completed
+  seq: 5
+  type: item.completed

server/packages/sandbox-agent/tests/server-manager/agent_server_manager.rs

@@ -0,0 +1,136 @@
+use std::sync::Arc;
+
+use sandbox_agent::router::test_utils::{exit_status, spawn_sleep_process, TestHarness};
+use sandbox_agent_agent_management::agents::AgentId;
+use sandbox_agent_universal_agent_schema::SessionEndReason;
+use tokio::time::{timeout, Duration};
+
+async fn wait_for_exit(child: &Arc<std::sync::Mutex<Option<std::process::Child>>>) {
+    for _ in 0..20 {
+        let done = {
+            let mut guard = child.lock().expect("child lock");
+            match guard.as_mut() {
+                Some(child) => child.try_wait().ok().flatten().is_some(),
+                None => true,
+            }
+        };
+        if done {
+            return;
+        }
+        tokio::time::sleep(Duration::from_millis(50)).await;
+    }
+}
+
+#[tokio::test]
+async fn register_and_unregister_sessions() {
+    let harness = TestHarness::new().await;
+    harness
+        .register_session(AgentId::Codex, "sess-1", Some("thread-1"))
+        .await;
+
+    assert!(
+        harness
+            .has_session_mapping(AgentId::Codex, "sess-1")
+            .await
+    );
+    assert_eq!(
+        harness
+            .native_mapping(AgentId::Codex, "thread-1")
+            .await
+            .as_deref(),
+        Some("sess-1")
+    );
+
+    harness
+        .unregister_session(AgentId::Codex, "sess-1", Some("thread-1"))
+        .await;
+
+    assert!(
+        !harness
+            .has_session_mapping(AgentId::Codex, "sess-1")
+            .await
+    );
+    assert!(
+        harness
+            .native_mapping(AgentId::Codex, "thread-1")
+            .await
+            .is_none()
+    );
+}
+
+#[tokio::test]
+async fn shutdown_marks_servers_stopped_and_kills_child() {
+    let harness = TestHarness::new().await;
+    let child = harness
+        .insert_stdio_server(AgentId::Codex, Some(spawn_sleep_process()), 0)
+        .await;
+
+    harness.shutdown().await;
+
+    assert!(matches!(
+        harness.server_status(AgentId::Codex).await,
+        Some(sandbox_agent::router::ServerStatus::Stopped)
+    ));
+
+    wait_for_exit(&child).await;
+    let exited = {
+        let mut guard = child.lock().expect("child lock");
+        guard
+            .as_mut()
+            .and_then(|child| child.try_wait().ok().flatten())
+            .is_some()
+    };
+    assert!(exited);
+}
+
+#[tokio::test]
+async fn handle_process_exit_marks_error_and_ends_sessions() {
+    let harness = TestHarness::new().await;
+    harness
+        .insert_session("sess-1", AgentId::Codex, Some("thread-1"))
+        .await;
+    harness
+        .register_session(AgentId::Codex, "sess-1", Some("thread-1"))
+        .await;
+    harness
+        .insert_stdio_server(AgentId::Codex, None, 1)
+        .await;
+
+    harness
+        .handle_process_exit(AgentId::Codex, 1, exit_status(7))
+        .await;
+
+    assert!(matches!(
+        harness.server_status(AgentId::Codex).await,
+        Some(sandbox_agent::router::ServerStatus::Error)
+    ));
+    assert!(
+        harness
+            .server_last_error(AgentId::Codex)
+            .await
+            .unwrap_or_default()
+            .contains("exited")
+    );
+    assert!(harness.session_ended("sess-1").await);
+    assert!(matches!(
+        harness.session_end_reason("sess-1").await,
+        Some(SessionEndReason::Error)
+    ));
+}
+
+#[tokio::test]
+async fn auto_restart_notifier_emits_signal() {
+    let harness = TestHarness::new().await;
+    let (tx, mut rx) = tokio::sync::mpsc::unbounded_channel();
+    harness.set_restart_notifier(tx).await;
+    harness.insert_http_server(AgentId::Mock, 2).await;
+
+    harness
+        .handle_process_exit(AgentId::Mock, 2, exit_status(2))
+        .await;
+
+    let received = timeout(Duration::from_millis(200), rx.recv())
+        .await
+        .expect("timeout");
+    assert_eq!(received, Some(AgentId::Mock));
+}

server/packages/sandbox-agent/tests/server-manager/mod.rs

@@ -0,0 +1 @@
+mod agent_server_manager;

server/packages/sandbox-agent/tests/server_manager.rs

@@ -0,0 +1,2 @@
+#[path = "server-manager/mod.rs"]
+mod server_manager;

server/packages/sandbox-agent/tests/ui.rs

@@ -0,0 +1,2 @@
+#[path = "ui/mod.rs"]
+mod ui;

server/packages/sandbox-agent/tests/ui/mod.rs

@@ -0,0 +1 @@
+mod inspector_ui;