mirror of
https://github.com/openai/codex.git
synced 2026-10-21 12:10:06 +00:00
Add free-form asynchronous user messages (#42354)
## What changed - Add the `send_message_to_user_async` control tool for concise questions and updates during ongoing work. - Emit messages as asynchronous agent items without ending the turn, and reject empty message text. - Expose the tool only to root agents when the model catalog explicitly opts in, independently of `request_user_input_async`. ## Testing - Cover catalog gating for root agents and subagents. - Verify asynchronous item emission, tool output, empty-message validation, and continued turn execution. GitOrigin-RevId: 8c6a7988a502f3dbc9eff7000913e85f0c596550
This commit is contained in:
@@ -24,6 +24,7 @@ pub(crate) mod request_plugin_install_spec;
|
||||
mod request_user_input;
|
||||
mod request_user_input_async;
|
||||
pub(crate) mod request_user_input_spec;
|
||||
mod send_message_to_user_async;
|
||||
pub(crate) mod shell_spec;
|
||||
mod sleep;
|
||||
mod test_sync;
|
||||
@@ -70,6 +71,7 @@ pub use request_permissions::RequestPermissionsHandler;
|
||||
pub use request_plugin_install::RequestPluginInstallHandler;
|
||||
pub use request_user_input::RequestUserInputHandler;
|
||||
pub use request_user_input_async::RequestUserInputAsyncHandler;
|
||||
pub use send_message_to_user_async::SendMessageToUserAsyncHandler;
|
||||
pub use sleep::SleepHandler;
|
||||
pub use test_sync::TestSyncHandler;
|
||||
pub(crate) use tool_search::ToolSearchHandlerCache;
|
||||
|
||||
@@ -0,0 +1,109 @@
|
||||
use crate::function_tool::FunctionCallError;
|
||||
use crate::tools::context::FunctionToolOutput;
|
||||
use crate::tools::context::ToolInvocation;
|
||||
use crate::tools::context::ToolPayload;
|
||||
use crate::tools::context::boxed_tool_output;
|
||||
use crate::tools::handlers::parse_arguments;
|
||||
use crate::tools::registry::CoreToolRuntime;
|
||||
use crate::tools::registry::ToolExecutor;
|
||||
use codex_protocol::items::AgentMessageContent;
|
||||
use codex_protocol::items::AgentMessageDelivery;
|
||||
use codex_protocol::items::AgentMessageItem;
|
||||
use codex_protocol::items::TurnItem;
|
||||
use codex_protocol::models::MessagePhase;
|
||||
use codex_tools::JsonSchema;
|
||||
use codex_tools::ResponsesApiTool;
|
||||
use codex_tools::ToolName;
|
||||
use codex_tools::ToolSpec;
|
||||
use serde::Deserialize;
|
||||
use std::collections::BTreeMap;
|
||||
|
||||
const TOOL_NAME: &str = "send_message_to_user_async";
|
||||
|
||||
pub struct SendMessageToUserAsyncHandler;
|
||||
|
||||
#[derive(Debug, Deserialize)]
|
||||
#[serde(deny_unknown_fields)]
|
||||
struct SendMessageToUserAsyncArgs {
|
||||
message: String,
|
||||
}
|
||||
|
||||
impl ToolExecutor<ToolInvocation> for SendMessageToUserAsyncHandler {
|
||||
fn tool_name(&self) -> ToolName {
|
||||
ToolName::plain(TOOL_NAME)
|
||||
}
|
||||
|
||||
fn spec(&self) -> ToolSpec {
|
||||
let properties = BTreeMap::from([(
|
||||
"message".to_string(),
|
||||
JsonSchema::string(Some(
|
||||
"The concise question or update to send to the user.".to_string(),
|
||||
)),
|
||||
)]);
|
||||
|
||||
ToolSpec::Function(ResponsesApiTool {
|
||||
name: TOOL_NAME.to_string(),
|
||||
description: "Send a concise message that needs the user's attention during ongoing work. The tool returns immediately without ending the turn or waiting for a reply; any reply arrives asynchronously as a new user message. Use this tool to ask for missing information, preferences, constraints, clarification, or approval; report a critical blocker or a finding that may change the task's direction; or answer a user question or status request received while work is still in progress. Use this tool when a message needs the user's immediate attention; use commentary for routine progress and intermediate context. Use clear formatting, such as bolding questions, to make requests easy to notice and answer."
|
||||
.to_string(),
|
||||
strict: false,
|
||||
defer_loading: None,
|
||||
parameters: JsonSchema::object(
|
||||
properties,
|
||||
Some(vec!["message".to_string()]),
|
||||
/*additional_properties*/ Some(false.into()),
|
||||
),
|
||||
output_schema: None,
|
||||
})
|
||||
}
|
||||
|
||||
fn handle<'a>(&'a self, invocation: ToolInvocation) -> codex_tools::ToolExecutorFuture<'a>
|
||||
where
|
||||
ToolInvocation: 'a,
|
||||
{
|
||||
Box::pin(async move {
|
||||
let ToolInvocation {
|
||||
session,
|
||||
turn,
|
||||
call_id,
|
||||
payload,
|
||||
..
|
||||
} = invocation;
|
||||
let ToolPayload::Function { arguments } = payload else {
|
||||
return Err(FunctionCallError::RespondToModel(format!(
|
||||
"{TOOL_NAME} handler received unsupported payload"
|
||||
)));
|
||||
};
|
||||
let args: SendMessageToUserAsyncArgs = parse_arguments(&arguments)?;
|
||||
let message = args.message.trim();
|
||||
if message.is_empty() {
|
||||
return Err(FunctionCallError::RespondToModel(
|
||||
"message must not be empty".to_string(),
|
||||
));
|
||||
}
|
||||
|
||||
let item = TurnItem::AgentMessage(AgentMessageItem {
|
||||
id: call_id,
|
||||
content: vec![AgentMessageContent::Text {
|
||||
text: message.to_string(),
|
||||
}],
|
||||
phase: Some(MessagePhase::FinalAnswer),
|
||||
memory_citation: None,
|
||||
delivery: Some(AgentMessageDelivery::Async),
|
||||
questions: None,
|
||||
});
|
||||
session.emit_turn_item_started(turn.as_ref(), &item).await;
|
||||
session.emit_turn_item_completed(turn.as_ref(), item).await;
|
||||
|
||||
Ok(boxed_tool_output(FunctionToolOutput::from_text(
|
||||
r#"{"accepted":true}"#.to_string(),
|
||||
/*success*/ Some(true),
|
||||
)))
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
impl CoreToolRuntime for SendMessageToUserAsyncHandler {
|
||||
fn is_builtin_control_tool(&self) -> bool {
|
||||
true
|
||||
}
|
||||
}
|
||||
@@ -25,6 +25,7 @@ use crate::tools::handlers::RequestPermissionsHandler;
|
||||
use crate::tools::handlers::RequestPluginInstallHandler;
|
||||
use crate::tools::handlers::RequestUserInputAsyncHandler;
|
||||
use crate::tools::handlers::RequestUserInputHandler;
|
||||
use crate::tools::handlers::SendMessageToUserAsyncHandler;
|
||||
use crate::tools::handlers::SleepHandler;
|
||||
use crate::tools::handlers::TestSyncHandler;
|
||||
use crate::tools::handlers::ToolSearchHandlerCache;
|
||||
@@ -1183,6 +1184,16 @@ fn add_core_utility_tools(context: &CoreToolPlanContext<'_>, registry: &mut Tool
|
||||
);
|
||||
}
|
||||
|
||||
if !turn_context.session_source.is_non_root_agent()
|
||||
&& context
|
||||
.model_info
|
||||
.experimental_supported_tools
|
||||
.iter()
|
||||
.any(|tool| tool == "send_message_to_user_async")
|
||||
{
|
||||
registry.add_with_exposure(SendMessageToUserAsyncHandler, ToolExposure::DirectModelOnly);
|
||||
}
|
||||
|
||||
if environment_mode.has_environment() && features.enabled(Feature::RequestPermissionsTool) {
|
||||
registry.add(RequestPermissionsHandler);
|
||||
}
|
||||
|
||||
@@ -120,6 +120,195 @@ async fn persistent_async_message_guidance_follows_tool_availability(
|
||||
Ok(())
|
||||
}
|
||||
|
||||
#[test_case(SessionSource::Exec, None, false; "root_without_opt_in")]
|
||||
#[test_case(SessionSource::Exec, Some("send_user_message_async"), false; "root_with_legacy_question_tool")]
|
||||
#[test_case(SessionSource::Exec, Some("request_user_input_async"), false; "root_with_question_tool")]
|
||||
#[test_case(SessionSource::Exec, Some("send_message_to_user_async"), true; "root_with_freeform_tool")]
|
||||
#[test_case(SessionSource::SubAgent(SubAgentSource::Other("test".to_string())), Some("send_message_to_user_async"), false; "subagent_with_freeform_tool")]
|
||||
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
||||
async fn freeform_async_message_requires_root_and_exact_catalog_opt_in(
|
||||
session_source: SessionSource,
|
||||
catalog_tool: Option<&'static str>,
|
||||
expected: bool,
|
||||
) -> Result<()> {
|
||||
skip_if_no_network!(Ok(()));
|
||||
|
||||
let server = start_mock_server().await;
|
||||
let responses = mount_sse_once(
|
||||
&server,
|
||||
sse(vec![ev_response_created("resp-1"), ev_completed("resp-1")]),
|
||||
)
|
||||
.await;
|
||||
let test = test_codex()
|
||||
.with_model_info_override("gpt-5.2", move |model| {
|
||||
model.tool_mode = Some(ToolMode::CodeModeOnly);
|
||||
model.experimental_supported_tools =
|
||||
catalog_tool.map(str::to_string).into_iter().collect();
|
||||
})
|
||||
.build_with_auto_env(&server)
|
||||
.await?;
|
||||
let thread = test
|
||||
.thread_manager
|
||||
.start_thread(StartThreadOptions {
|
||||
session_source: Some(session_source),
|
||||
..StartThreadOptions::new(test.config.clone())
|
||||
})
|
||||
.await?
|
||||
.thread;
|
||||
thread
|
||||
.start_or_steer_turn(TurnInputRequest::user_input(vec![UserInput::Text {
|
||||
text: "Report progress.".to_string(),
|
||||
text_elements: Vec::new(),
|
||||
}]))
|
||||
.await?;
|
||||
wait_for_event(&thread, |event| matches!(event, EventMsg::TurnComplete(_))).await;
|
||||
|
||||
let body = responses.single_request().body_json();
|
||||
let tools = body["tools"].as_array().expect("request tools");
|
||||
assert_eq!(
|
||||
tools
|
||||
.iter()
|
||||
.any(|tool| tool["name"] == "send_message_to_user_async"),
|
||||
expected,
|
||||
);
|
||||
Ok(())
|
||||
}
|
||||
|
||||
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
||||
async fn freeform_async_message_emits_an_item_without_ending_the_turn() -> Result<()> {
|
||||
skip_if_no_network!(Ok(()));
|
||||
|
||||
const CALL_ID: &str = "freeform-message";
|
||||
const MESSAGE: &str = "I found a blocker that changes the plan.";
|
||||
let server = start_mock_server().await;
|
||||
let responses = mount_sse_sequence(
|
||||
&server,
|
||||
vec![
|
||||
sse(vec![
|
||||
ev_response_created("resp-1"),
|
||||
ev_function_call_with_namespace(
|
||||
CALL_ID,
|
||||
"functions",
|
||||
"send_message_to_user_async",
|
||||
&json!({ "message": format!(" {MESSAGE} ") }).to_string(),
|
||||
),
|
||||
ev_completed("resp-1"),
|
||||
]),
|
||||
sse(vec![
|
||||
ev_response_created("resp-2"),
|
||||
ev_function_call_with_namespace(
|
||||
"blank-message",
|
||||
"functions",
|
||||
"send_message_to_user_async",
|
||||
&json!({ "message": " " }).to_string(),
|
||||
),
|
||||
ev_completed("resp-2"),
|
||||
]),
|
||||
sse(vec![
|
||||
ev_response_created("resp-3"),
|
||||
ev_assistant_message("final-message", "Finished."),
|
||||
ev_completed("resp-3"),
|
||||
]),
|
||||
],
|
||||
)
|
||||
.await;
|
||||
let test = test_codex()
|
||||
.with_model_info_override("gpt-5.2", |model| {
|
||||
model.tool_mode = Some(ToolMode::CodeModeOnly);
|
||||
model.experimental_supported_tools.extend([
|
||||
"send_message_to_user_async".to_string(),
|
||||
"request_user_input_async".to_string(),
|
||||
]);
|
||||
model
|
||||
.model_messages
|
||||
.as_mut()
|
||||
.expect("model instruction metadata")
|
||||
.tools = Some(ToolMessages {
|
||||
send_user_message_async: Some(ToolMessage {
|
||||
description: Some("Questions only.".to_string()),
|
||||
}),
|
||||
});
|
||||
})
|
||||
.build_with_auto_env(&server)
|
||||
.await?;
|
||||
test.codex
|
||||
.start_or_steer_turn(TurnInputRequest::user_input(vec![UserInput::Text {
|
||||
text: "Keep me updated.".to_string(),
|
||||
text_elements: Vec::new(),
|
||||
}]))
|
||||
.await?;
|
||||
|
||||
let started = wait_for_event_match(test.codex.as_ref(), |event| match event {
|
||||
EventMsg::ItemStarted(event) if event.item.id() == CALL_ID => Some(event.item.clone()),
|
||||
_ => None,
|
||||
})
|
||||
.await;
|
||||
let completed = wait_for_event_match(test.codex.as_ref(), |event| match event {
|
||||
EventMsg::ItemCompleted(event) if event.item.id() == CALL_ID => Some(event.item.clone()),
|
||||
_ => None,
|
||||
})
|
||||
.await;
|
||||
let expected = serde_json::to_value(TurnItem::AgentMessage(AgentMessageItem {
|
||||
id: CALL_ID.to_string(),
|
||||
content: vec![AgentMessageContent::Text {
|
||||
text: MESSAGE.to_string(),
|
||||
}],
|
||||
phase: Some(MessagePhase::FinalAnswer),
|
||||
memory_citation: None,
|
||||
delivery: Some(AgentMessageDelivery::Async),
|
||||
questions: None,
|
||||
}))?;
|
||||
assert_eq!(serde_json::to_value(started)?, expected);
|
||||
assert_eq!(serde_json::to_value(completed)?, expected);
|
||||
wait_for_event(test.codex.as_ref(), |event| {
|
||||
let item = match event {
|
||||
EventMsg::ItemStarted(event) => Some(&event.item),
|
||||
EventMsg::ItemCompleted(event) => Some(&event.item),
|
||||
_ => None,
|
||||
};
|
||||
assert!(!matches!(item, Some(TurnItem::AgentMessage(message)) if message.delivery == Some(AgentMessageDelivery::Async)));
|
||||
matches!(event, EventMsg::TurnComplete(_))
|
||||
}).await;
|
||||
|
||||
let requests = responses.requests();
|
||||
assert_eq!(requests.len(), 3);
|
||||
let body = requests[0].body_json();
|
||||
let tools = body["tools"].as_array().expect("request tools");
|
||||
let freeform_tool = tools
|
||||
.iter()
|
||||
.find(|tool| tool["name"] == "send_message_to_user_async")
|
||||
.expect("freeform tool is directly available");
|
||||
assert_eq!(freeform_tool["parameters"]["required"], json!(["message"]));
|
||||
assert!(
|
||||
freeform_tool["description"]
|
||||
.as_str()
|
||||
.expect("description")
|
||||
.contains("report a critical blocker")
|
||||
);
|
||||
let question_tool = tools
|
||||
.iter()
|
||||
.find(|tool| tool["name"] == "request_user_input_async")
|
||||
.expect("question tool is still directly available");
|
||||
assert_eq!(question_tool["description"], "Questions only.");
|
||||
assert_eq!(
|
||||
requests[1].function_call_output_text(CALL_ID),
|
||||
Some(r#"{"accepted":true}"#.to_string())
|
||||
);
|
||||
assert_eq!(
|
||||
requests[2].function_call_output_text("blank-message"),
|
||||
Some("message must not be empty".to_string())
|
||||
);
|
||||
assert!(
|
||||
!requests[1]
|
||||
.input()
|
||||
.into_iter()
|
||||
.any(|item| item["type"] == "message"
|
||||
&& item["role"] == "assistant"
|
||||
&& item.to_string().contains(MESSAGE))
|
||||
);
|
||||
Ok(())
|
||||
}
|
||||
|
||||
#[test_case(None, "send_user_message_async"; "fallback_description")]
|
||||
#[test_case(None, "request_user_input_async"; "current_catalog_name")]
|
||||
#[test_case(Some(ToolMessages { send_user_message_async: None }), "send_user_message_async"; "missing_tool")]
|
||||
|
||||
Reference in New Issue
Block a user