1use super::*;
2use acp_thread::{AgentConnection, AgentModelGroupName, AgentModelList, UserMessageId};
3use action_log::ActionLog;
4use agent_client_protocol::{self as acp};
5use agent_settings::AgentProfileId;
6use anyhow::Result;
7use client::{Client, UserStore};
8use fs::{FakeFs, Fs};
9use futures::{StreamExt, channel::mpsc::UnboundedReceiver};
10use gpui::{
11 App, AppContext, Entity, Task, TestAppContext, UpdateGlobal, http_client::FakeHttpClient,
12};
13use indoc::indoc;
14use language_model::{
15 LanguageModel, LanguageModelCompletionError, LanguageModelCompletionEvent, LanguageModelId,
16 LanguageModelProviderName, LanguageModelRegistry, LanguageModelRequestMessage,
17 LanguageModelToolResult, LanguageModelToolUse, MessageContent, Role, StopReason,
18 fake_provider::FakeLanguageModel,
19};
20use pretty_assertions::assert_eq;
21use project::Project;
22use prompt_store::ProjectContext;
23use reqwest_client::ReqwestClient;
24use schemars::JsonSchema;
25use serde::{Deserialize, Serialize};
26use serde_json::json;
27use settings::SettingsStore;
28use std::{path::Path, rc::Rc, sync::Arc, time::Duration};
29use util::path;
30
31mod test_tools;
32use test_tools::*;
33
34#[gpui::test]
35#[ignore = "can't run on CI yet"]
36async fn test_echo(cx: &mut TestAppContext) {
37 let ThreadTest { thread, .. } = setup(cx, TestModel::Sonnet4).await;
38
39 let events = thread
40 .update(cx, |thread, cx| {
41 thread.send(UserMessageId::new(), ["Testing: Reply with 'Hello'"], cx)
42 })
43 .unwrap()
44 .collect()
45 .await;
46 thread.update(cx, |thread, _cx| {
47 assert_eq!(
48 thread.last_message().unwrap().to_markdown(),
49 indoc! {"
50 ## Assistant
51
52 Hello
53 "}
54 )
55 });
56 assert_eq!(stop_events(events), vec![acp::StopReason::EndTurn]);
57}
58
59#[gpui::test]
60#[ignore = "can't run on CI yet"]
61async fn test_thinking(cx: &mut TestAppContext) {
62 let ThreadTest { thread, .. } = setup(cx, TestModel::Sonnet4Thinking).await;
63
64 let events = thread
65 .update(cx, |thread, cx| {
66 thread.send(
67 UserMessageId::new(),
68 [indoc! {"
69 Testing:
70
71 Generate a thinking step where you just think the word 'Think',
72 and have your final answer be 'Hello'
73 "}],
74 cx,
75 )
76 })
77 .unwrap()
78 .collect()
79 .await;
80 thread.update(cx, |thread, _cx| {
81 assert_eq!(
82 thread.last_message().unwrap().to_markdown(),
83 indoc! {"
84 ## Assistant
85
86 <think>Think</think>
87 Hello
88 "}
89 )
90 });
91 assert_eq!(stop_events(events), vec![acp::StopReason::EndTurn]);
92}
93
94#[gpui::test]
95async fn test_system_prompt(cx: &mut TestAppContext) {
96 let ThreadTest {
97 model,
98 thread,
99 project_context,
100 ..
101 } = setup(cx, TestModel::Fake).await;
102 let fake_model = model.as_fake();
103
104 project_context.update(cx, |project_context, _cx| {
105 project_context.shell = "test-shell".into()
106 });
107 thread.update(cx, |thread, _| thread.add_tool(EchoTool));
108 thread
109 .update(cx, |thread, cx| {
110 thread.send(UserMessageId::new(), ["abc"], cx)
111 })
112 .unwrap();
113 cx.run_until_parked();
114 let mut pending_completions = fake_model.pending_completions();
115 assert_eq!(
116 pending_completions.len(),
117 1,
118 "unexpected pending completions: {:?}",
119 pending_completions
120 );
121
122 let pending_completion = pending_completions.pop().unwrap();
123 assert_eq!(pending_completion.messages[0].role, Role::System);
124
125 let system_message = &pending_completion.messages[0];
126 let system_prompt = system_message.content[0].to_str().unwrap();
127 assert!(
128 system_prompt.contains("test-shell"),
129 "unexpected system message: {:?}",
130 system_message
131 );
132 assert!(
133 system_prompt.contains("## Fixing Diagnostics"),
134 "unexpected system message: {:?}",
135 system_message
136 );
137}
138
139#[gpui::test]
140async fn test_prompt_caching(cx: &mut TestAppContext) {
141 let ThreadTest { model, thread, .. } = setup(cx, TestModel::Fake).await;
142 let fake_model = model.as_fake();
143
144 // Send initial user message and verify it's cached
145 thread
146 .update(cx, |thread, cx| {
147 thread.send(UserMessageId::new(), ["Message 1"], cx)
148 })
149 .unwrap();
150 cx.run_until_parked();
151
152 let completion = fake_model.pending_completions().pop().unwrap();
153 assert_eq!(
154 completion.messages[1..],
155 vec![LanguageModelRequestMessage {
156 role: Role::User,
157 content: vec!["Message 1".into()],
158 cache: true
159 }]
160 );
161 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::Text(
162 "Response to Message 1".into(),
163 ));
164 fake_model.end_last_completion_stream();
165 cx.run_until_parked();
166
167 // Send another user message and verify only the latest is cached
168 thread
169 .update(cx, |thread, cx| {
170 thread.send(UserMessageId::new(), ["Message 2"], cx)
171 })
172 .unwrap();
173 cx.run_until_parked();
174
175 let completion = fake_model.pending_completions().pop().unwrap();
176 assert_eq!(
177 completion.messages[1..],
178 vec![
179 LanguageModelRequestMessage {
180 role: Role::User,
181 content: vec!["Message 1".into()],
182 cache: false
183 },
184 LanguageModelRequestMessage {
185 role: Role::Assistant,
186 content: vec!["Response to Message 1".into()],
187 cache: false
188 },
189 LanguageModelRequestMessage {
190 role: Role::User,
191 content: vec!["Message 2".into()],
192 cache: true
193 }
194 ]
195 );
196 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::Text(
197 "Response to Message 2".into(),
198 ));
199 fake_model.end_last_completion_stream();
200 cx.run_until_parked();
201
202 // Simulate a tool call and verify that the latest tool result is cached
203 thread.update(cx, |thread, _| thread.add_tool(EchoTool));
204 thread
205 .update(cx, |thread, cx| {
206 thread.send(UserMessageId::new(), ["Use the echo tool"], cx)
207 })
208 .unwrap();
209 cx.run_until_parked();
210
211 let tool_use = LanguageModelToolUse {
212 id: "tool_1".into(),
213 name: EchoTool.name().into(),
214 raw_input: json!({"text": "test"}).to_string(),
215 input: json!({"text": "test"}),
216 is_input_complete: true,
217 };
218 fake_model
219 .send_last_completion_stream_event(LanguageModelCompletionEvent::ToolUse(tool_use.clone()));
220 fake_model.end_last_completion_stream();
221 cx.run_until_parked();
222
223 let completion = fake_model.pending_completions().pop().unwrap();
224 let tool_result = LanguageModelToolResult {
225 tool_use_id: "tool_1".into(),
226 tool_name: EchoTool.name().into(),
227 is_error: false,
228 content: "test".into(),
229 output: Some("test".into()),
230 };
231 assert_eq!(
232 completion.messages[1..],
233 vec![
234 LanguageModelRequestMessage {
235 role: Role::User,
236 content: vec!["Message 1".into()],
237 cache: false
238 },
239 LanguageModelRequestMessage {
240 role: Role::Assistant,
241 content: vec!["Response to Message 1".into()],
242 cache: false
243 },
244 LanguageModelRequestMessage {
245 role: Role::User,
246 content: vec!["Message 2".into()],
247 cache: false
248 },
249 LanguageModelRequestMessage {
250 role: Role::Assistant,
251 content: vec!["Response to Message 2".into()],
252 cache: false
253 },
254 LanguageModelRequestMessage {
255 role: Role::User,
256 content: vec!["Use the echo tool".into()],
257 cache: false
258 },
259 LanguageModelRequestMessage {
260 role: Role::Assistant,
261 content: vec![MessageContent::ToolUse(tool_use)],
262 cache: false
263 },
264 LanguageModelRequestMessage {
265 role: Role::User,
266 content: vec![MessageContent::ToolResult(tool_result)],
267 cache: true
268 }
269 ]
270 );
271}
272
273#[gpui::test]
274#[ignore = "can't run on CI yet"]
275async fn test_basic_tool_calls(cx: &mut TestAppContext) {
276 let ThreadTest { thread, .. } = setup(cx, TestModel::Sonnet4).await;
277
278 // Test a tool call that's likely to complete *before* streaming stops.
279 let events = thread
280 .update(cx, |thread, cx| {
281 thread.add_tool(EchoTool);
282 thread.send(
283 UserMessageId::new(),
284 ["Now test the echo tool with 'Hello'. Does it work? Say 'Yes' or 'No'."],
285 cx,
286 )
287 })
288 .unwrap()
289 .collect()
290 .await;
291 assert_eq!(stop_events(events), vec![acp::StopReason::EndTurn]);
292
293 // Test a tool calls that's likely to complete *after* streaming stops.
294 let events = thread
295 .update(cx, |thread, cx| {
296 thread.remove_tool(&AgentTool::name(&EchoTool));
297 thread.add_tool(DelayTool);
298 thread.send(
299 UserMessageId::new(),
300 [
301 "Now call the delay tool with 200ms.",
302 "When the timer goes off, then you echo the output of the tool.",
303 ],
304 cx,
305 )
306 })
307 .unwrap()
308 .collect()
309 .await;
310 assert_eq!(stop_events(events), vec![acp::StopReason::EndTurn]);
311 thread.update(cx, |thread, _cx| {
312 assert!(
313 thread
314 .last_message()
315 .unwrap()
316 .as_agent_message()
317 .unwrap()
318 .content
319 .iter()
320 .any(|content| {
321 if let AgentMessageContent::Text(text) = content {
322 text.contains("Ding")
323 } else {
324 false
325 }
326 }),
327 "{}",
328 thread.to_markdown()
329 );
330 });
331}
332
333#[gpui::test]
334#[ignore = "can't run on CI yet"]
335async fn test_streaming_tool_calls(cx: &mut TestAppContext) {
336 let ThreadTest { thread, .. } = setup(cx, TestModel::Sonnet4).await;
337
338 // Test a tool call that's likely to complete *before* streaming stops.
339 let mut events = thread
340 .update(cx, |thread, cx| {
341 thread.add_tool(WordListTool);
342 thread.send(UserMessageId::new(), ["Test the word_list tool."], cx)
343 })
344 .unwrap();
345
346 let mut saw_partial_tool_use = false;
347 while let Some(event) = events.next().await {
348 if let Ok(AgentResponseEvent::ToolCall(tool_call)) = event {
349 thread.update(cx, |thread, _cx| {
350 // Look for a tool use in the thread's last message
351 let message = thread.last_message().unwrap();
352 let agent_message = message.as_agent_message().unwrap();
353 let last_content = agent_message.content.last().unwrap();
354 if let AgentMessageContent::ToolUse(last_tool_use) = last_content {
355 assert_eq!(last_tool_use.name.as_ref(), "word_list");
356 if tool_call.status == acp::ToolCallStatus::Pending {
357 if !last_tool_use.is_input_complete
358 && last_tool_use.input.get("g").is_none()
359 {
360 saw_partial_tool_use = true;
361 }
362 } else {
363 last_tool_use
364 .input
365 .get("a")
366 .expect("'a' has streamed because input is now complete");
367 last_tool_use
368 .input
369 .get("g")
370 .expect("'g' has streamed because input is now complete");
371 }
372 } else {
373 panic!("last content should be a tool use");
374 }
375 });
376 }
377 }
378
379 assert!(
380 saw_partial_tool_use,
381 "should see at least one partially streamed tool use in the history"
382 );
383}
384
385#[gpui::test]
386async fn test_tool_authorization(cx: &mut TestAppContext) {
387 let ThreadTest { model, thread, .. } = setup(cx, TestModel::Fake).await;
388 let fake_model = model.as_fake();
389
390 let mut events = thread
391 .update(cx, |thread, cx| {
392 thread.add_tool(ToolRequiringPermission);
393 thread.send(UserMessageId::new(), ["abc"], cx)
394 })
395 .unwrap();
396 cx.run_until_parked();
397 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::ToolUse(
398 LanguageModelToolUse {
399 id: "tool_id_1".into(),
400 name: ToolRequiringPermission.name().into(),
401 raw_input: "{}".into(),
402 input: json!({}),
403 is_input_complete: true,
404 },
405 ));
406 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::ToolUse(
407 LanguageModelToolUse {
408 id: "tool_id_2".into(),
409 name: ToolRequiringPermission.name().into(),
410 raw_input: "{}".into(),
411 input: json!({}),
412 is_input_complete: true,
413 },
414 ));
415 fake_model.end_last_completion_stream();
416 let tool_call_auth_1 = next_tool_call_authorization(&mut events).await;
417 let tool_call_auth_2 = next_tool_call_authorization(&mut events).await;
418
419 // Approve the first
420 tool_call_auth_1
421 .response
422 .send(tool_call_auth_1.options[1].id.clone())
423 .unwrap();
424 cx.run_until_parked();
425
426 // Reject the second
427 tool_call_auth_2
428 .response
429 .send(tool_call_auth_1.options[2].id.clone())
430 .unwrap();
431 cx.run_until_parked();
432
433 let completion = fake_model.pending_completions().pop().unwrap();
434 let message = completion.messages.last().unwrap();
435 assert_eq!(
436 message.content,
437 vec![
438 language_model::MessageContent::ToolResult(LanguageModelToolResult {
439 tool_use_id: tool_call_auth_1.tool_call.id.0.to_string().into(),
440 tool_name: ToolRequiringPermission.name().into(),
441 is_error: false,
442 content: "Allowed".into(),
443 output: Some("Allowed".into())
444 }),
445 language_model::MessageContent::ToolResult(LanguageModelToolResult {
446 tool_use_id: tool_call_auth_2.tool_call.id.0.to_string().into(),
447 tool_name: ToolRequiringPermission.name().into(),
448 is_error: true,
449 content: "Permission to run tool denied by user".into(),
450 output: None
451 })
452 ]
453 );
454
455 // Simulate yet another tool call.
456 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::ToolUse(
457 LanguageModelToolUse {
458 id: "tool_id_3".into(),
459 name: ToolRequiringPermission.name().into(),
460 raw_input: "{}".into(),
461 input: json!({}),
462 is_input_complete: true,
463 },
464 ));
465 fake_model.end_last_completion_stream();
466
467 // Respond by always allowing tools.
468 let tool_call_auth_3 = next_tool_call_authorization(&mut events).await;
469 tool_call_auth_3
470 .response
471 .send(tool_call_auth_3.options[0].id.clone())
472 .unwrap();
473 cx.run_until_parked();
474 let completion = fake_model.pending_completions().pop().unwrap();
475 let message = completion.messages.last().unwrap();
476 assert_eq!(
477 message.content,
478 vec![language_model::MessageContent::ToolResult(
479 LanguageModelToolResult {
480 tool_use_id: tool_call_auth_3.tool_call.id.0.to_string().into(),
481 tool_name: ToolRequiringPermission.name().into(),
482 is_error: false,
483 content: "Allowed".into(),
484 output: Some("Allowed".into())
485 }
486 )]
487 );
488
489 // Simulate a final tool call, ensuring we don't trigger authorization.
490 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::ToolUse(
491 LanguageModelToolUse {
492 id: "tool_id_4".into(),
493 name: ToolRequiringPermission.name().into(),
494 raw_input: "{}".into(),
495 input: json!({}),
496 is_input_complete: true,
497 },
498 ));
499 fake_model.end_last_completion_stream();
500 cx.run_until_parked();
501 let completion = fake_model.pending_completions().pop().unwrap();
502 let message = completion.messages.last().unwrap();
503 assert_eq!(
504 message.content,
505 vec![language_model::MessageContent::ToolResult(
506 LanguageModelToolResult {
507 tool_use_id: "tool_id_4".into(),
508 tool_name: ToolRequiringPermission.name().into(),
509 is_error: false,
510 content: "Allowed".into(),
511 output: Some("Allowed".into())
512 }
513 )]
514 );
515}
516
517#[gpui::test]
518async fn test_tool_hallucination(cx: &mut TestAppContext) {
519 let ThreadTest { model, thread, .. } = setup(cx, TestModel::Fake).await;
520 let fake_model = model.as_fake();
521
522 let mut events = thread
523 .update(cx, |thread, cx| {
524 thread.send(UserMessageId::new(), ["abc"], cx)
525 })
526 .unwrap();
527 cx.run_until_parked();
528 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::ToolUse(
529 LanguageModelToolUse {
530 id: "tool_id_1".into(),
531 name: "nonexistent_tool".into(),
532 raw_input: "{}".into(),
533 input: json!({}),
534 is_input_complete: true,
535 },
536 ));
537 fake_model.end_last_completion_stream();
538
539 let tool_call = expect_tool_call(&mut events).await;
540 assert_eq!(tool_call.title, "nonexistent_tool");
541 assert_eq!(tool_call.status, acp::ToolCallStatus::Pending);
542 let update = expect_tool_call_update_fields(&mut events).await;
543 assert_eq!(update.fields.status, Some(acp::ToolCallStatus::Failed));
544}
545
546#[gpui::test]
547async fn test_resume_after_tool_use_limit(cx: &mut TestAppContext) {
548 let ThreadTest { model, thread, .. } = setup(cx, TestModel::Fake).await;
549 let fake_model = model.as_fake();
550
551 let events = thread
552 .update(cx, |thread, cx| {
553 thread.add_tool(EchoTool);
554 thread.send(UserMessageId::new(), ["abc"], cx)
555 })
556 .unwrap();
557 cx.run_until_parked();
558 let tool_use = LanguageModelToolUse {
559 id: "tool_id_1".into(),
560 name: EchoTool.name().into(),
561 raw_input: "{}".into(),
562 input: serde_json::to_value(&EchoToolInput { text: "def".into() }).unwrap(),
563 is_input_complete: true,
564 };
565 fake_model
566 .send_last_completion_stream_event(LanguageModelCompletionEvent::ToolUse(tool_use.clone()));
567 fake_model.end_last_completion_stream();
568
569 cx.run_until_parked();
570 let completion = fake_model.pending_completions().pop().unwrap();
571 let tool_result = LanguageModelToolResult {
572 tool_use_id: "tool_id_1".into(),
573 tool_name: EchoTool.name().into(),
574 is_error: false,
575 content: "def".into(),
576 output: Some("def".into()),
577 };
578 assert_eq!(
579 completion.messages[1..],
580 vec![
581 LanguageModelRequestMessage {
582 role: Role::User,
583 content: vec!["abc".into()],
584 cache: false
585 },
586 LanguageModelRequestMessage {
587 role: Role::Assistant,
588 content: vec![MessageContent::ToolUse(tool_use.clone())],
589 cache: false
590 },
591 LanguageModelRequestMessage {
592 role: Role::User,
593 content: vec![MessageContent::ToolResult(tool_result.clone())],
594 cache: true
595 },
596 ]
597 );
598
599 // Simulate reaching tool use limit.
600 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::StatusUpdate(
601 cloud_llm_client::CompletionRequestStatus::ToolUseLimitReached,
602 ));
603 fake_model.end_last_completion_stream();
604 let last_event = events.collect::<Vec<_>>().await.pop().unwrap();
605 assert!(
606 last_event
607 .unwrap_err()
608 .is::<language_model::ToolUseLimitReachedError>()
609 );
610
611 let events = thread.update(cx, |thread, cx| thread.resume(cx)).unwrap();
612 cx.run_until_parked();
613 let completion = fake_model.pending_completions().pop().unwrap();
614 assert_eq!(
615 completion.messages[1..],
616 vec![
617 LanguageModelRequestMessage {
618 role: Role::User,
619 content: vec!["abc".into()],
620 cache: false
621 },
622 LanguageModelRequestMessage {
623 role: Role::Assistant,
624 content: vec![MessageContent::ToolUse(tool_use)],
625 cache: false
626 },
627 LanguageModelRequestMessage {
628 role: Role::User,
629 content: vec![MessageContent::ToolResult(tool_result)],
630 cache: false
631 },
632 LanguageModelRequestMessage {
633 role: Role::User,
634 content: vec!["Continue where you left off".into()],
635 cache: true
636 }
637 ]
638 );
639
640 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::Text("Done".into()));
641 fake_model.end_last_completion_stream();
642 events.collect::<Vec<_>>().await;
643 thread.read_with(cx, |thread, _cx| {
644 assert_eq!(
645 thread.last_message().unwrap().to_markdown(),
646 indoc! {"
647 ## Assistant
648
649 Done
650 "}
651 )
652 });
653
654 // Ensure we error if calling resume when tool use limit was *not* reached.
655 let error = thread
656 .update(cx, |thread, cx| thread.resume(cx))
657 .unwrap_err();
658 assert_eq!(
659 error.to_string(),
660 "can only resume after tool use limit is reached"
661 )
662}
663
664#[gpui::test]
665async fn test_send_after_tool_use_limit(cx: &mut TestAppContext) {
666 let ThreadTest { model, thread, .. } = setup(cx, TestModel::Fake).await;
667 let fake_model = model.as_fake();
668
669 let events = thread
670 .update(cx, |thread, cx| {
671 thread.add_tool(EchoTool);
672 thread.send(UserMessageId::new(), ["abc"], cx)
673 })
674 .unwrap();
675 cx.run_until_parked();
676
677 let tool_use = LanguageModelToolUse {
678 id: "tool_id_1".into(),
679 name: EchoTool.name().into(),
680 raw_input: "{}".into(),
681 input: serde_json::to_value(&EchoToolInput { text: "def".into() }).unwrap(),
682 is_input_complete: true,
683 };
684 let tool_result = LanguageModelToolResult {
685 tool_use_id: "tool_id_1".into(),
686 tool_name: EchoTool.name().into(),
687 is_error: false,
688 content: "def".into(),
689 output: Some("def".into()),
690 };
691 fake_model
692 .send_last_completion_stream_event(LanguageModelCompletionEvent::ToolUse(tool_use.clone()));
693 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::StatusUpdate(
694 cloud_llm_client::CompletionRequestStatus::ToolUseLimitReached,
695 ));
696 fake_model.end_last_completion_stream();
697 let last_event = events.collect::<Vec<_>>().await.pop().unwrap();
698 assert!(
699 last_event
700 .unwrap_err()
701 .is::<language_model::ToolUseLimitReachedError>()
702 );
703
704 thread
705 .update(cx, |thread, cx| {
706 thread.send(UserMessageId::new(), vec!["ghi"], cx)
707 })
708 .unwrap();
709 cx.run_until_parked();
710 let completion = fake_model.pending_completions().pop().unwrap();
711 assert_eq!(
712 completion.messages[1..],
713 vec![
714 LanguageModelRequestMessage {
715 role: Role::User,
716 content: vec!["abc".into()],
717 cache: false
718 },
719 LanguageModelRequestMessage {
720 role: Role::Assistant,
721 content: vec![MessageContent::ToolUse(tool_use)],
722 cache: false
723 },
724 LanguageModelRequestMessage {
725 role: Role::User,
726 content: vec![MessageContent::ToolResult(tool_result)],
727 cache: false
728 },
729 LanguageModelRequestMessage {
730 role: Role::User,
731 content: vec!["ghi".into()],
732 cache: true
733 }
734 ]
735 );
736}
737
738async fn expect_tool_call(
739 events: &mut UnboundedReceiver<Result<AgentResponseEvent>>,
740) -> acp::ToolCall {
741 let event = events
742 .next()
743 .await
744 .expect("no tool call authorization event received")
745 .unwrap();
746 match event {
747 AgentResponseEvent::ToolCall(tool_call) => return tool_call,
748 event => {
749 panic!("Unexpected event {event:?}");
750 }
751 }
752}
753
754async fn expect_tool_call_update_fields(
755 events: &mut UnboundedReceiver<Result<AgentResponseEvent>>,
756) -> acp::ToolCallUpdate {
757 let event = events
758 .next()
759 .await
760 .expect("no tool call authorization event received")
761 .unwrap();
762 match event {
763 AgentResponseEvent::ToolCallUpdate(acp_thread::ToolCallUpdate::UpdateFields(update)) => {
764 return update;
765 }
766 event => {
767 panic!("Unexpected event {event:?}");
768 }
769 }
770}
771
772async fn next_tool_call_authorization(
773 events: &mut UnboundedReceiver<Result<AgentResponseEvent>>,
774) -> ToolCallAuthorization {
775 loop {
776 let event = events
777 .next()
778 .await
779 .expect("no tool call authorization event received")
780 .unwrap();
781 if let AgentResponseEvent::ToolCallAuthorization(tool_call_authorization) = event {
782 let permission_kinds = tool_call_authorization
783 .options
784 .iter()
785 .map(|o| o.kind)
786 .collect::<Vec<_>>();
787 assert_eq!(
788 permission_kinds,
789 vec![
790 acp::PermissionOptionKind::AllowAlways,
791 acp::PermissionOptionKind::AllowOnce,
792 acp::PermissionOptionKind::RejectOnce,
793 ]
794 );
795 return tool_call_authorization;
796 }
797 }
798}
799
800#[gpui::test]
801#[ignore = "can't run on CI yet"]
802async fn test_concurrent_tool_calls(cx: &mut TestAppContext) {
803 let ThreadTest { thread, .. } = setup(cx, TestModel::Sonnet4).await;
804
805 // Test concurrent tool calls with different delay times
806 let events = thread
807 .update(cx, |thread, cx| {
808 thread.add_tool(DelayTool);
809 thread.send(
810 UserMessageId::new(),
811 [
812 "Call the delay tool twice in the same message.",
813 "Once with 100ms. Once with 300ms.",
814 "When both timers are complete, describe the outputs.",
815 ],
816 cx,
817 )
818 })
819 .unwrap()
820 .collect()
821 .await;
822
823 let stop_reasons = stop_events(events);
824 assert_eq!(stop_reasons, vec![acp::StopReason::EndTurn]);
825
826 thread.update(cx, |thread, _cx| {
827 let last_message = thread.last_message().unwrap();
828 let agent_message = last_message.as_agent_message().unwrap();
829 let text = agent_message
830 .content
831 .iter()
832 .filter_map(|content| {
833 if let AgentMessageContent::Text(text) = content {
834 Some(text.as_str())
835 } else {
836 None
837 }
838 })
839 .collect::<String>();
840
841 assert!(text.contains("Ding"));
842 });
843}
844
845#[gpui::test]
846async fn test_profiles(cx: &mut TestAppContext) {
847 let ThreadTest {
848 model, thread, fs, ..
849 } = setup(cx, TestModel::Fake).await;
850 let fake_model = model.as_fake();
851
852 thread.update(cx, |thread, _cx| {
853 thread.add_tool(DelayTool);
854 thread.add_tool(EchoTool);
855 thread.add_tool(InfiniteTool);
856 });
857
858 // Override profiles and wait for settings to be loaded.
859 fs.insert_file(
860 paths::settings_file(),
861 json!({
862 "agent": {
863 "profiles": {
864 "test-1": {
865 "name": "Test Profile 1",
866 "tools": {
867 EchoTool.name(): true,
868 DelayTool.name(): true,
869 }
870 },
871 "test-2": {
872 "name": "Test Profile 2",
873 "tools": {
874 InfiniteTool.name(): true,
875 }
876 }
877 }
878 }
879 })
880 .to_string()
881 .into_bytes(),
882 )
883 .await;
884 cx.run_until_parked();
885
886 // Test that test-1 profile (default) has echo and delay tools
887 thread
888 .update(cx, |thread, cx| {
889 thread.set_profile(AgentProfileId("test-1".into()));
890 thread.send(UserMessageId::new(), ["test"], cx)
891 })
892 .unwrap();
893 cx.run_until_parked();
894
895 let mut pending_completions = fake_model.pending_completions();
896 assert_eq!(pending_completions.len(), 1);
897 let completion = pending_completions.pop().unwrap();
898 let tool_names: Vec<String> = completion
899 .tools
900 .iter()
901 .map(|tool| tool.name.clone())
902 .collect();
903 assert_eq!(tool_names, vec![DelayTool.name(), EchoTool.name()]);
904 fake_model.end_last_completion_stream();
905
906 // Switch to test-2 profile, and verify that it has only the infinite tool.
907 thread
908 .update(cx, |thread, cx| {
909 thread.set_profile(AgentProfileId("test-2".into()));
910 thread.send(UserMessageId::new(), ["test2"], cx)
911 })
912 .unwrap();
913 cx.run_until_parked();
914 let mut pending_completions = fake_model.pending_completions();
915 assert_eq!(pending_completions.len(), 1);
916 let completion = pending_completions.pop().unwrap();
917 let tool_names: Vec<String> = completion
918 .tools
919 .iter()
920 .map(|tool| tool.name.clone())
921 .collect();
922 assert_eq!(tool_names, vec![InfiniteTool.name()]);
923}
924
925#[gpui::test]
926#[ignore = "can't run on CI yet"]
927async fn test_cancellation(cx: &mut TestAppContext) {
928 let ThreadTest { thread, .. } = setup(cx, TestModel::Sonnet4).await;
929
930 let mut events = thread
931 .update(cx, |thread, cx| {
932 thread.add_tool(InfiniteTool);
933 thread.add_tool(EchoTool);
934 thread.send(
935 UserMessageId::new(),
936 ["Call the echo tool, then call the infinite tool, then explain their output"],
937 cx,
938 )
939 })
940 .unwrap();
941
942 // Wait until both tools are called.
943 let mut expected_tools = vec!["Echo", "Infinite Tool"];
944 let mut echo_id = None;
945 let mut echo_completed = false;
946 while let Some(event) = events.next().await {
947 match event.unwrap() {
948 AgentResponseEvent::ToolCall(tool_call) => {
949 assert_eq!(tool_call.title, expected_tools.remove(0));
950 if tool_call.title == "Echo" {
951 echo_id = Some(tool_call.id);
952 }
953 }
954 AgentResponseEvent::ToolCallUpdate(acp_thread::ToolCallUpdate::UpdateFields(
955 acp::ToolCallUpdate {
956 id,
957 fields:
958 acp::ToolCallUpdateFields {
959 status: Some(acp::ToolCallStatus::Completed),
960 ..
961 },
962 },
963 )) if Some(&id) == echo_id.as_ref() => {
964 echo_completed = true;
965 }
966 _ => {}
967 }
968
969 if expected_tools.is_empty() && echo_completed {
970 break;
971 }
972 }
973
974 // Cancel the current send and ensure that the event stream is closed, even
975 // if one of the tools is still running.
976 thread.update(cx, |thread, _cx| thread.cancel());
977 let events = events.collect::<Vec<_>>().await;
978 let last_event = events.last();
979 assert!(
980 matches!(
981 last_event,
982 Some(Ok(AgentResponseEvent::Stop(acp::StopReason::Canceled)))
983 ),
984 "unexpected event {last_event:?}"
985 );
986
987 // Ensure we can still send a new message after cancellation.
988 let events = thread
989 .update(cx, |thread, cx| {
990 thread.send(
991 UserMessageId::new(),
992 ["Testing: reply with 'Hello' then stop."],
993 cx,
994 )
995 })
996 .unwrap()
997 .collect::<Vec<_>>()
998 .await;
999 thread.update(cx, |thread, _cx| {
1000 let message = thread.last_message().unwrap();
1001 let agent_message = message.as_agent_message().unwrap();
1002 assert_eq!(
1003 agent_message.content,
1004 vec![AgentMessageContent::Text("Hello".to_string())]
1005 );
1006 });
1007 assert_eq!(stop_events(events), vec![acp::StopReason::EndTurn]);
1008}
1009
1010#[gpui::test]
1011async fn test_in_progress_send_canceled_by_next_send(cx: &mut TestAppContext) {
1012 let ThreadTest { model, thread, .. } = setup(cx, TestModel::Fake).await;
1013 let fake_model = model.as_fake();
1014
1015 let events_1 = thread
1016 .update(cx, |thread, cx| {
1017 thread.send(UserMessageId::new(), ["Hello 1"], cx)
1018 })
1019 .unwrap();
1020 cx.run_until_parked();
1021 fake_model.send_last_completion_stream_text_chunk("Hey 1!");
1022 cx.run_until_parked();
1023
1024 let events_2 = thread
1025 .update(cx, |thread, cx| {
1026 thread.send(UserMessageId::new(), ["Hello 2"], cx)
1027 })
1028 .unwrap();
1029 cx.run_until_parked();
1030 fake_model.send_last_completion_stream_text_chunk("Hey 2!");
1031 fake_model
1032 .send_last_completion_stream_event(LanguageModelCompletionEvent::Stop(StopReason::EndTurn));
1033 fake_model.end_last_completion_stream();
1034
1035 let events_1 = events_1.collect::<Vec<_>>().await;
1036 assert_eq!(stop_events(events_1), vec![acp::StopReason::Canceled]);
1037 let events_2 = events_2.collect::<Vec<_>>().await;
1038 assert_eq!(stop_events(events_2), vec![acp::StopReason::EndTurn]);
1039}
1040
1041#[gpui::test]
1042async fn test_subsequent_successful_sends_dont_cancel(cx: &mut TestAppContext) {
1043 let ThreadTest { model, thread, .. } = setup(cx, TestModel::Fake).await;
1044 let fake_model = model.as_fake();
1045
1046 let events_1 = thread
1047 .update(cx, |thread, cx| {
1048 thread.send(UserMessageId::new(), ["Hello 1"], cx)
1049 })
1050 .unwrap();
1051 cx.run_until_parked();
1052 fake_model.send_last_completion_stream_text_chunk("Hey 1!");
1053 fake_model
1054 .send_last_completion_stream_event(LanguageModelCompletionEvent::Stop(StopReason::EndTurn));
1055 fake_model.end_last_completion_stream();
1056 let events_1 = events_1.collect::<Vec<_>>().await;
1057
1058 let events_2 = thread
1059 .update(cx, |thread, cx| {
1060 thread.send(UserMessageId::new(), ["Hello 2"], cx)
1061 })
1062 .unwrap();
1063 cx.run_until_parked();
1064 fake_model.send_last_completion_stream_text_chunk("Hey 2!");
1065 fake_model
1066 .send_last_completion_stream_event(LanguageModelCompletionEvent::Stop(StopReason::EndTurn));
1067 fake_model.end_last_completion_stream();
1068 let events_2 = events_2.collect::<Vec<_>>().await;
1069
1070 assert_eq!(stop_events(events_1), vec![acp::StopReason::EndTurn]);
1071 assert_eq!(stop_events(events_2), vec![acp::StopReason::EndTurn]);
1072}
1073
1074#[gpui::test]
1075async fn test_refusal(cx: &mut TestAppContext) {
1076 let ThreadTest { model, thread, .. } = setup(cx, TestModel::Fake).await;
1077 let fake_model = model.as_fake();
1078
1079 let events = thread
1080 .update(cx, |thread, cx| {
1081 thread.send(UserMessageId::new(), ["Hello"], cx)
1082 })
1083 .unwrap();
1084 cx.run_until_parked();
1085 thread.read_with(cx, |thread, _| {
1086 assert_eq!(
1087 thread.to_markdown(),
1088 indoc! {"
1089 ## User
1090
1091 Hello
1092 "}
1093 );
1094 });
1095
1096 fake_model.send_last_completion_stream_text_chunk("Hey!");
1097 cx.run_until_parked();
1098 thread.read_with(cx, |thread, _| {
1099 assert_eq!(
1100 thread.to_markdown(),
1101 indoc! {"
1102 ## User
1103
1104 Hello
1105
1106 ## Assistant
1107
1108 Hey!
1109 "}
1110 );
1111 });
1112
1113 // If the model refuses to continue, the thread should remove all the messages after the last user message.
1114 fake_model
1115 .send_last_completion_stream_event(LanguageModelCompletionEvent::Stop(StopReason::Refusal));
1116 let events = events.collect::<Vec<_>>().await;
1117 assert_eq!(stop_events(events), vec![acp::StopReason::Refusal]);
1118 thread.read_with(cx, |thread, _| {
1119 assert_eq!(thread.to_markdown(), "");
1120 });
1121}
1122
1123#[gpui::test]
1124async fn test_truncate(cx: &mut TestAppContext) {
1125 let ThreadTest { model, thread, .. } = setup(cx, TestModel::Fake).await;
1126 let fake_model = model.as_fake();
1127
1128 let message_id = UserMessageId::new();
1129 thread
1130 .update(cx, |thread, cx| {
1131 thread.send(message_id.clone(), ["Hello"], cx)
1132 })
1133 .unwrap();
1134 cx.run_until_parked();
1135 thread.read_with(cx, |thread, _| {
1136 assert_eq!(
1137 thread.to_markdown(),
1138 indoc! {"
1139 ## User
1140
1141 Hello
1142 "}
1143 );
1144 });
1145
1146 fake_model.send_last_completion_stream_text_chunk("Hey!");
1147 cx.run_until_parked();
1148 thread.read_with(cx, |thread, _| {
1149 assert_eq!(
1150 thread.to_markdown(),
1151 indoc! {"
1152 ## User
1153
1154 Hello
1155
1156 ## Assistant
1157
1158 Hey!
1159 "}
1160 );
1161 });
1162
1163 thread
1164 .update(cx, |thread, _cx| thread.truncate(message_id))
1165 .unwrap();
1166 cx.run_until_parked();
1167 thread.read_with(cx, |thread, _| {
1168 assert_eq!(thread.to_markdown(), "");
1169 });
1170
1171 // Ensure we can still send a new message after truncation.
1172 thread
1173 .update(cx, |thread, cx| {
1174 thread.send(UserMessageId::new(), ["Hi"], cx)
1175 })
1176 .unwrap();
1177 thread.update(cx, |thread, _cx| {
1178 assert_eq!(
1179 thread.to_markdown(),
1180 indoc! {"
1181 ## User
1182
1183 Hi
1184 "}
1185 );
1186 });
1187 cx.run_until_parked();
1188 fake_model.send_last_completion_stream_text_chunk("Ahoy!");
1189 cx.run_until_parked();
1190 thread.read_with(cx, |thread, _| {
1191 assert_eq!(
1192 thread.to_markdown(),
1193 indoc! {"
1194 ## User
1195
1196 Hi
1197
1198 ## Assistant
1199
1200 Ahoy!
1201 "}
1202 );
1203 });
1204}
1205
1206#[gpui::test]
1207async fn test_agent_connection(cx: &mut TestAppContext) {
1208 cx.update(settings::init);
1209 let templates = Templates::new();
1210
1211 // Initialize language model system with test provider
1212 cx.update(|cx| {
1213 gpui_tokio::init(cx);
1214 client::init_settings(cx);
1215
1216 let http_client = FakeHttpClient::with_404_response();
1217 let clock = Arc::new(clock::FakeSystemClock::new());
1218 let client = Client::new(clock, http_client, cx);
1219 let user_store = cx.new(|cx| UserStore::new(client.clone(), cx));
1220 language_model::init(client.clone(), cx);
1221 language_models::init(user_store.clone(), client.clone(), cx);
1222 Project::init_settings(cx);
1223 LanguageModelRegistry::test(cx);
1224 agent_settings::init(cx);
1225 });
1226 cx.executor().forbid_parking();
1227
1228 // Create a project for new_thread
1229 let fake_fs = cx.update(|cx| fs::FakeFs::new(cx.background_executor().clone()));
1230 fake_fs.insert_tree(path!("/test"), json!({})).await;
1231 let project = Project::test(fake_fs.clone(), [Path::new("/test")], cx).await;
1232 let cwd = Path::new("/test");
1233
1234 // Create agent and connection
1235 let agent = NativeAgent::new(
1236 project.clone(),
1237 templates.clone(),
1238 None,
1239 fake_fs.clone(),
1240 &mut cx.to_async(),
1241 )
1242 .await
1243 .unwrap();
1244 let connection = NativeAgentConnection(agent.clone());
1245
1246 // Test model_selector returns Some
1247 let selector_opt = connection.model_selector();
1248 assert!(
1249 selector_opt.is_some(),
1250 "agent2 should always support ModelSelector"
1251 );
1252 let selector = selector_opt.unwrap();
1253
1254 // Test list_models
1255 let listed_models = cx
1256 .update(|cx| selector.list_models(cx))
1257 .await
1258 .expect("list_models should succeed");
1259 let AgentModelList::Grouped(listed_models) = listed_models else {
1260 panic!("Unexpected model list type");
1261 };
1262 assert!(!listed_models.is_empty(), "should have at least one model");
1263 assert_eq!(
1264 listed_models[&AgentModelGroupName("Fake".into())][0].id.0,
1265 "fake/fake"
1266 );
1267
1268 // Create a thread using new_thread
1269 let connection_rc = Rc::new(connection.clone());
1270 let acp_thread = cx
1271 .update(|cx| connection_rc.new_thread(project, cwd, cx))
1272 .await
1273 .expect("new_thread should succeed");
1274
1275 // Get the session_id from the AcpThread
1276 let session_id = acp_thread.read_with(cx, |thread, _| thread.session_id().clone());
1277
1278 // Test selected_model returns the default
1279 let model = cx
1280 .update(|cx| selector.selected_model(&session_id, cx))
1281 .await
1282 .expect("selected_model should succeed");
1283 let model = cx
1284 .update(|cx| agent.read(cx).models().model_from_id(&model.id))
1285 .unwrap();
1286 let model = model.as_fake();
1287 assert_eq!(model.id().0, "fake", "should return default model");
1288
1289 let request = acp_thread.update(cx, |thread, cx| thread.send(vec!["abc".into()], cx));
1290 cx.run_until_parked();
1291 model.send_last_completion_stream_text_chunk("def");
1292 cx.run_until_parked();
1293 acp_thread.read_with(cx, |thread, cx| {
1294 assert_eq!(
1295 thread.to_markdown(cx),
1296 indoc! {"
1297 ## User
1298
1299 abc
1300
1301 ## Assistant
1302
1303 def
1304
1305 "}
1306 )
1307 });
1308
1309 // Test cancel
1310 cx.update(|cx| connection.cancel(&session_id, cx));
1311 request.await.expect("prompt should fail gracefully");
1312
1313 // Ensure that dropping the ACP thread causes the native thread to be
1314 // dropped as well.
1315 cx.update(|_| drop(acp_thread));
1316 let result = cx
1317 .update(|cx| {
1318 connection.prompt(
1319 Some(acp_thread::UserMessageId::new()),
1320 acp::PromptRequest {
1321 session_id: session_id.clone(),
1322 prompt: vec!["ghi".into()],
1323 },
1324 cx,
1325 )
1326 })
1327 .await;
1328 assert_eq!(
1329 result.as_ref().unwrap_err().to_string(),
1330 "Session not found",
1331 "unexpected result: {:?}",
1332 result
1333 );
1334}
1335
1336#[gpui::test]
1337async fn test_tool_updates_to_completion(cx: &mut TestAppContext) {
1338 let ThreadTest { thread, model, .. } = setup(cx, TestModel::Fake).await;
1339 thread.update(cx, |thread, _cx| thread.add_tool(ThinkingTool));
1340 let fake_model = model.as_fake();
1341
1342 let mut events = thread
1343 .update(cx, |thread, cx| {
1344 thread.send(UserMessageId::new(), ["Think"], cx)
1345 })
1346 .unwrap();
1347 cx.run_until_parked();
1348
1349 // Simulate streaming partial input.
1350 let input = json!({});
1351 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::ToolUse(
1352 LanguageModelToolUse {
1353 id: "1".into(),
1354 name: ThinkingTool.name().into(),
1355 raw_input: input.to_string(),
1356 input,
1357 is_input_complete: false,
1358 },
1359 ));
1360
1361 // Input streaming completed
1362 let input = json!({ "content": "Thinking hard!" });
1363 fake_model.send_last_completion_stream_event(LanguageModelCompletionEvent::ToolUse(
1364 LanguageModelToolUse {
1365 id: "1".into(),
1366 name: "thinking".into(),
1367 raw_input: input.to_string(),
1368 input,
1369 is_input_complete: true,
1370 },
1371 ));
1372 fake_model.end_last_completion_stream();
1373 cx.run_until_parked();
1374
1375 let tool_call = expect_tool_call(&mut events).await;
1376 assert_eq!(
1377 tool_call,
1378 acp::ToolCall {
1379 id: acp::ToolCallId("1".into()),
1380 title: "Thinking".into(),
1381 kind: acp::ToolKind::Think,
1382 status: acp::ToolCallStatus::Pending,
1383 content: vec![],
1384 locations: vec![],
1385 raw_input: Some(json!({})),
1386 raw_output: None,
1387 }
1388 );
1389 let update = expect_tool_call_update_fields(&mut events).await;
1390 assert_eq!(
1391 update,
1392 acp::ToolCallUpdate {
1393 id: acp::ToolCallId("1".into()),
1394 fields: acp::ToolCallUpdateFields {
1395 title: Some("Thinking".into()),
1396 kind: Some(acp::ToolKind::Think),
1397 raw_input: Some(json!({ "content": "Thinking hard!" })),
1398 ..Default::default()
1399 },
1400 }
1401 );
1402 let update = expect_tool_call_update_fields(&mut events).await;
1403 assert_eq!(
1404 update,
1405 acp::ToolCallUpdate {
1406 id: acp::ToolCallId("1".into()),
1407 fields: acp::ToolCallUpdateFields {
1408 status: Some(acp::ToolCallStatus::InProgress),
1409 ..Default::default()
1410 },
1411 }
1412 );
1413 let update = expect_tool_call_update_fields(&mut events).await;
1414 assert_eq!(
1415 update,
1416 acp::ToolCallUpdate {
1417 id: acp::ToolCallId("1".into()),
1418 fields: acp::ToolCallUpdateFields {
1419 content: Some(vec!["Thinking hard!".into()]),
1420 ..Default::default()
1421 },
1422 }
1423 );
1424 let update = expect_tool_call_update_fields(&mut events).await;
1425 assert_eq!(
1426 update,
1427 acp::ToolCallUpdate {
1428 id: acp::ToolCallId("1".into()),
1429 fields: acp::ToolCallUpdateFields {
1430 status: Some(acp::ToolCallStatus::Completed),
1431 raw_output: Some("Finished thinking.".into()),
1432 ..Default::default()
1433 },
1434 }
1435 );
1436}
1437
1438#[gpui::test]
1439async fn test_send_no_retry_on_success(cx: &mut TestAppContext) {
1440 let ThreadTest { thread, model, .. } = setup(cx, TestModel::Fake).await;
1441 let fake_model = model.as_fake();
1442
1443 let mut events = thread
1444 .update(cx, |thread, cx| {
1445 thread.set_completion_mode(agent_settings::CompletionMode::Burn);
1446 thread.send(UserMessageId::new(), ["Hello!"], cx)
1447 })
1448 .unwrap();
1449 cx.run_until_parked();
1450
1451 fake_model.send_last_completion_stream_text_chunk("Hey!");
1452 fake_model.end_last_completion_stream();
1453
1454 let mut retry_events = Vec::new();
1455 while let Some(Ok(event)) = events.next().await {
1456 match event {
1457 AgentResponseEvent::Retry(retry_status) => {
1458 retry_events.push(retry_status);
1459 }
1460 AgentResponseEvent::Stop(..) => break,
1461 _ => {}
1462 }
1463 }
1464
1465 assert_eq!(retry_events.len(), 0);
1466 thread.read_with(cx, |thread, _cx| {
1467 assert_eq!(
1468 thread.to_markdown(),
1469 indoc! {"
1470 ## User
1471
1472 Hello!
1473
1474 ## Assistant
1475
1476 Hey!
1477 "}
1478 )
1479 });
1480}
1481
1482#[gpui::test]
1483async fn test_send_retry_on_error(cx: &mut TestAppContext) {
1484 let ThreadTest { thread, model, .. } = setup(cx, TestModel::Fake).await;
1485 let fake_model = model.as_fake();
1486
1487 let mut events = thread
1488 .update(cx, |thread, cx| {
1489 thread.set_completion_mode(agent_settings::CompletionMode::Burn);
1490 thread.send(UserMessageId::new(), ["Hello!"], cx)
1491 })
1492 .unwrap();
1493 cx.run_until_parked();
1494
1495 fake_model.send_last_completion_stream_error(LanguageModelCompletionError::ServerOverloaded {
1496 provider: LanguageModelProviderName::new("Anthropic"),
1497 retry_after: Some(Duration::from_secs(3)),
1498 });
1499 fake_model.end_last_completion_stream();
1500
1501 cx.executor().advance_clock(Duration::from_secs(3));
1502 cx.run_until_parked();
1503
1504 fake_model.send_last_completion_stream_text_chunk("Hey!");
1505 fake_model.end_last_completion_stream();
1506
1507 let mut retry_events = Vec::new();
1508 while let Some(Ok(event)) = events.next().await {
1509 match event {
1510 AgentResponseEvent::Retry(retry_status) => {
1511 retry_events.push(retry_status);
1512 }
1513 AgentResponseEvent::Stop(..) => break,
1514 _ => {}
1515 }
1516 }
1517
1518 assert_eq!(retry_events.len(), 1);
1519 assert!(matches!(
1520 retry_events[0],
1521 acp_thread::RetryStatus { attempt: 1, .. }
1522 ));
1523 thread.read_with(cx, |thread, _cx| {
1524 assert_eq!(
1525 thread.to_markdown(),
1526 indoc! {"
1527 ## User
1528
1529 Hello!
1530
1531 ## Assistant
1532
1533 Hey!
1534 "}
1535 )
1536 });
1537}
1538
1539#[gpui::test]
1540async fn test_send_max_retries_exceeded(cx: &mut TestAppContext) {
1541 let ThreadTest { thread, model, .. } = setup(cx, TestModel::Fake).await;
1542 let fake_model = model.as_fake();
1543
1544 let mut events = thread
1545 .update(cx, |thread, cx| {
1546 thread.set_completion_mode(agent_settings::CompletionMode::Burn);
1547 thread.send(UserMessageId::new(), ["Hello!"], cx)
1548 })
1549 .unwrap();
1550 cx.run_until_parked();
1551
1552 for _ in 0..crate::thread::MAX_RETRY_ATTEMPTS + 1 {
1553 fake_model.send_last_completion_stream_error(
1554 LanguageModelCompletionError::ServerOverloaded {
1555 provider: LanguageModelProviderName::new("Anthropic"),
1556 retry_after: Some(Duration::from_secs(3)),
1557 },
1558 );
1559 fake_model.end_last_completion_stream();
1560 cx.executor().advance_clock(Duration::from_secs(3));
1561 cx.run_until_parked();
1562 }
1563
1564 let mut errors = Vec::new();
1565 let mut retry_events = Vec::new();
1566 while let Some(event) = events.next().await {
1567 match event {
1568 Ok(AgentResponseEvent::Retry(retry_status)) => {
1569 retry_events.push(retry_status);
1570 }
1571 Ok(AgentResponseEvent::Stop(..)) => break,
1572 Err(error) => errors.push(error),
1573 _ => {}
1574 }
1575 }
1576
1577 assert_eq!(
1578 retry_events.len(),
1579 crate::thread::MAX_RETRY_ATTEMPTS as usize
1580 );
1581 for i in 0..crate::thread::MAX_RETRY_ATTEMPTS as usize {
1582 assert_eq!(retry_events[i].attempt, i + 1);
1583 }
1584 assert_eq!(errors.len(), 1);
1585 let error = errors[0]
1586 .downcast_ref::<LanguageModelCompletionError>()
1587 .unwrap();
1588 assert!(matches!(
1589 error,
1590 LanguageModelCompletionError::ServerOverloaded { .. }
1591 ));
1592}
1593
1594/// Filters out the stop events for asserting against in tests
1595fn stop_events(result_events: Vec<Result<AgentResponseEvent>>) -> Vec<acp::StopReason> {
1596 result_events
1597 .into_iter()
1598 .filter_map(|event| match event.unwrap() {
1599 AgentResponseEvent::Stop(stop_reason) => Some(stop_reason),
1600 _ => None,
1601 })
1602 .collect()
1603}
1604
1605struct ThreadTest {
1606 model: Arc<dyn LanguageModel>,
1607 thread: Entity<Thread>,
1608 project_context: Entity<ProjectContext>,
1609 fs: Arc<FakeFs>,
1610}
1611
1612enum TestModel {
1613 Sonnet4,
1614 Sonnet4Thinking,
1615 Fake,
1616}
1617
1618impl TestModel {
1619 fn id(&self) -> LanguageModelId {
1620 match self {
1621 TestModel::Sonnet4 => LanguageModelId("claude-sonnet-4-latest".into()),
1622 TestModel::Sonnet4Thinking => LanguageModelId("claude-sonnet-4-thinking-latest".into()),
1623 TestModel::Fake => unreachable!(),
1624 }
1625 }
1626}
1627
1628async fn setup(cx: &mut TestAppContext, model: TestModel) -> ThreadTest {
1629 cx.executor().allow_parking();
1630
1631 let fs = FakeFs::new(cx.background_executor.clone());
1632 fs.create_dir(paths::settings_file().parent().unwrap())
1633 .await
1634 .unwrap();
1635 fs.insert_file(
1636 paths::settings_file(),
1637 json!({
1638 "agent": {
1639 "default_profile": "test-profile",
1640 "profiles": {
1641 "test-profile": {
1642 "name": "Test Profile",
1643 "tools": {
1644 EchoTool.name(): true,
1645 DelayTool.name(): true,
1646 WordListTool.name(): true,
1647 ToolRequiringPermission.name(): true,
1648 InfiniteTool.name(): true,
1649 }
1650 }
1651 }
1652 }
1653 })
1654 .to_string()
1655 .into_bytes(),
1656 )
1657 .await;
1658
1659 cx.update(|cx| {
1660 settings::init(cx);
1661 Project::init_settings(cx);
1662 agent_settings::init(cx);
1663 gpui_tokio::init(cx);
1664 let http_client = ReqwestClient::user_agent("agent tests").unwrap();
1665 cx.set_http_client(Arc::new(http_client));
1666
1667 client::init_settings(cx);
1668 let client = Client::production(cx);
1669 let user_store = cx.new(|cx| UserStore::new(client.clone(), cx));
1670 language_model::init(client.clone(), cx);
1671 language_models::init(user_store.clone(), client.clone(), cx);
1672
1673 watch_settings(fs.clone(), cx);
1674 });
1675
1676 let templates = Templates::new();
1677
1678 fs.insert_tree(path!("/test"), json!({})).await;
1679 let project = Project::test(fs.clone(), [path!("/test").as_ref()], cx).await;
1680
1681 let model = cx
1682 .update(|cx| {
1683 if let TestModel::Fake = model {
1684 Task::ready(Arc::new(FakeLanguageModel::default()) as Arc<_>)
1685 } else {
1686 let model_id = model.id();
1687 let models = LanguageModelRegistry::read_global(cx);
1688 let model = models
1689 .available_models(cx)
1690 .find(|model| model.id() == model_id)
1691 .unwrap();
1692
1693 let provider = models.provider(&model.provider_id()).unwrap();
1694 let authenticated = provider.authenticate(cx);
1695
1696 cx.spawn(async move |_cx| {
1697 authenticated.await.unwrap();
1698 model
1699 })
1700 }
1701 })
1702 .await;
1703
1704 let project_context = cx.new(|_cx| ProjectContext::default());
1705 let context_server_registry =
1706 cx.new(|cx| ContextServerRegistry::new(project.read(cx).context_server_store(), cx));
1707 let action_log = cx.new(|_| ActionLog::new(project.clone()));
1708 let thread = cx.new(|cx| {
1709 Thread::new(
1710 project,
1711 project_context.clone(),
1712 context_server_registry,
1713 action_log,
1714 templates,
1715 Some(model.clone()),
1716 cx,
1717 )
1718 });
1719 ThreadTest {
1720 model,
1721 thread,
1722 project_context,
1723 fs,
1724 }
1725}
1726
1727#[cfg(test)]
1728#[ctor::ctor]
1729fn init_logger() {
1730 if std::env::var("RUST_LOG").is_ok() {
1731 env_logger::init();
1732 }
1733}
1734
1735fn watch_settings(fs: Arc<dyn Fs>, cx: &mut App) {
1736 let fs = fs.clone();
1737 cx.spawn({
1738 async move |cx| {
1739 let mut new_settings_content_rx = settings::watch_config_file(
1740 cx.background_executor(),
1741 fs,
1742 paths::settings_file().clone(),
1743 );
1744
1745 while let Some(new_settings_content) = new_settings_content_rx.next().await {
1746 cx.update(|cx| {
1747 SettingsStore::update_global(cx, |settings, cx| {
1748 settings.set_user_settings(&new_settings_content, cx)
1749 })
1750 })
1751 .ok();
1752 }
1753 }
1754 })
1755 .detach();
1756}