mirror of
https://github.com/openai/codex.git
synced 2026-04-24 22:54:54 +00:00
bdc4742bfc
### Motivation - Allow MCP OAuth flows to request scopes defined in `config.toml` instead of requiring users to always pass `--scopes` on the CLI. CLI/remote parameters should still override config values. ### Description - Add optional `scopes: Option<Vec<String>>` to `McpServerConfig` and `RawMcpServerConfig`, and propagate it through deserialization and the built config types. - Serialize `scopes` into the MCP server TOML via `serialize_mcp_server_table` in `core/src/config/edit.rs` and include `scopes` in the generated config schema (`core/config.schema.json`). - CLI: update `codex-rs/cli/src/mcp_cmd.rs` `run_login` to fall back to `server.scopes` when the `--scopes` flag is empty, with explicit CLI scopes still taking precedence. - App server: update `codex-rs/app-server/src/codex_message_processor.rs` `mcp_server_oauth_login` to use `params.scopes.or_else(|| server.scopes.clone())` so the RPC path also respects configured scopes. - Update many test fixtures to initialize the new `scopes` field (set to `None`) so test code builds with the new struct field. ### Testing - Ran config tooling and formatters: `just write-config-schema` (succeeded), `just fmt` (succeeded), and `just fix -p codex-core`, `just fix -p codex-cli`, `just fix -p codex-app-server` (succeeded where applicable). - Ran unit tests for the CLI: `cargo test -p codex-cli` (passed). - Ran unit tests for core: `cargo test -p codex-core` (ran; many tests passed but several failed, including model refresh/403-related tests, shell snapshot/timeouts, and several `unified_exec` expectations). - Ran app-server tests: `cargo test -p codex-app-server` (ran; many integration-suite tests failed due to mocked/remote HTTP 401/403 responses and wiremock expectations). If you want, I can split the tests into smaller focused runs or help debug the failing integration tests (they appear to be unrelated to the config change and stem from external HTTP/mocking behaviors encountered during the test runs). ------ [Codex Task](https://chatgpt.com/codex/tasks/task_i_69718f505914832ea1f334b3ba064553)
836 lines
27 KiB
Rust
836 lines
27 KiB
Rust
#![cfg(not(target_os = "windows"))]
|
|
#![allow(clippy::unwrap_used, clippy::expect_used)]
|
|
|
|
use anyhow::Context;
|
|
use anyhow::Result;
|
|
use codex_core::config::types::McpServerConfig;
|
|
use codex_core::config::types::McpServerTransportConfig;
|
|
use codex_core::protocol::AskForApproval;
|
|
use codex_core::protocol::EventMsg;
|
|
use codex_core::protocol::Op;
|
|
use codex_core::protocol::SandboxPolicy;
|
|
use codex_protocol::config_types::ReasoningSummary;
|
|
use codex_protocol::user_input::UserInput;
|
|
use core_test_support::assert_regex_match;
|
|
use core_test_support::responses;
|
|
use core_test_support::responses::ev_assistant_message;
|
|
use core_test_support::responses::ev_completed;
|
|
use core_test_support::responses::ev_function_call;
|
|
use core_test_support::responses::ev_response_created;
|
|
use core_test_support::responses::mount_sse_once;
|
|
use core_test_support::responses::mount_sse_sequence;
|
|
use core_test_support::responses::sse;
|
|
use core_test_support::responses::start_mock_server;
|
|
use core_test_support::skip_if_no_network;
|
|
use core_test_support::stdio_server_bin;
|
|
use core_test_support::test_codex::test_codex;
|
|
use core_test_support::wait_for_event;
|
|
use serde_json::Value;
|
|
use serde_json::json;
|
|
use std::collections::HashMap;
|
|
use std::time::Duration;
|
|
|
|
// Verifies byte-truncation formatting for function error output (RespondToModel errors)
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn truncate_function_error_trims_respond_to_model() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
let mut builder = test_codex().with_model("test-gpt-5.1-codex");
|
|
let test = builder.build(&server).await?;
|
|
|
|
// Construct a very long, non-existent path to force a RespondToModel error with a large message
|
|
let long_path = "long path text should trigger truncation".repeat(8_000);
|
|
let call_id = "grep-huge-error";
|
|
let args = json!({
|
|
"pattern": "alpha",
|
|
"path": long_path,
|
|
"limit": 10
|
|
});
|
|
let responses = vec![
|
|
sse(vec![
|
|
ev_response_created("resp-1"),
|
|
ev_function_call(call_id, "grep_files", &serde_json::to_string(&args)?),
|
|
ev_completed("resp-1"),
|
|
]),
|
|
sse(vec![
|
|
ev_assistant_message("msg-1", "done"),
|
|
ev_completed("resp-2"),
|
|
]),
|
|
];
|
|
let mock = mount_sse_sequence(&server, responses).await;
|
|
|
|
test.submit_turn_with_policy(
|
|
"trigger grep_files with long path to test truncation",
|
|
SandboxPolicy::DangerFullAccess,
|
|
)
|
|
.await?;
|
|
|
|
let output = mock
|
|
.function_call_output_text(call_id)
|
|
.context("function error output present")?;
|
|
|
|
tracing::debug!(output = %output, "truncated function error output");
|
|
|
|
// Expect plaintext with token-based truncation marker and no omitted-lines marker
|
|
assert!(
|
|
serde_json::from_str::<serde_json::Value>(&output).is_err(),
|
|
"expected error output to be plain text",
|
|
);
|
|
assert!(
|
|
!output.contains("Total output lines:"),
|
|
"error output should not include line-based truncation header: {output}",
|
|
);
|
|
let truncated_pattern = r"(?s)^unable to access `.*tokens truncated.*$";
|
|
assert_regex_match(truncated_pattern, &output);
|
|
assert!(
|
|
!output.contains("omitted"),
|
|
"line omission marker should not appear when no lines were dropped: {output}"
|
|
);
|
|
|
|
Ok(())
|
|
}
|
|
|
|
// Verifies that a standard tool call (shell_command) exceeding the model formatting
|
|
// limits is truncated before being sent back to the model.
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn tool_call_output_configured_limit_chars_type() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
|
|
// Use a model that exposes the shell_command tool.
|
|
let mut builder = test_codex().with_model("gpt-5.1").with_config(|config| {
|
|
config.tool_output_token_limit = Some(100_000);
|
|
});
|
|
|
|
let fixture = builder.build(&server).await?;
|
|
|
|
let call_id = "shell-too-large";
|
|
let command = if cfg!(windows) {
|
|
"for ($i=1; $i -le 100000; $i++) { Write-Output $i }"
|
|
} else {
|
|
"seq 1 100000"
|
|
};
|
|
let args = serde_json::json!({
|
|
"command": command,
|
|
"timeout_ms": 5_000,
|
|
});
|
|
|
|
// First response: model tells us to run the tool; second: complete the turn.
|
|
mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_response_created("resp-1"),
|
|
responses::ev_function_call(call_id, "shell_command", &serde_json::to_string(&args)?),
|
|
responses::ev_completed("resp-1"),
|
|
]),
|
|
)
|
|
.await;
|
|
let mock2 = mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_assistant_message("msg-1", "done"),
|
|
responses::ev_completed("resp-2"),
|
|
]),
|
|
)
|
|
.await;
|
|
|
|
fixture
|
|
.submit_turn_with_policy("trigger big shell output", SandboxPolicy::DangerFullAccess)
|
|
.await?;
|
|
|
|
// Inspect what we sent back to the model; it should contain a truncated
|
|
// function_call_output for the shell call.
|
|
let output = mock2
|
|
.single_request()
|
|
.function_call_output_text(call_id)
|
|
.context("function_call_output present for shell call")?;
|
|
let output = output.replace("\r\n", "\n");
|
|
|
|
// Expect plain text (not JSON) containing the entire shell output.
|
|
assert!(
|
|
serde_json::from_str::<Value>(&output).is_err(),
|
|
"expected truncated shell output to be plain text"
|
|
);
|
|
|
|
assert!(
|
|
(400000..=401000).contains(&output.len()),
|
|
"we should be almost 100k tokens"
|
|
);
|
|
|
|
assert!(
|
|
!output.contains("tokens truncated"),
|
|
"shell output should not contain tokens truncated marker: {output}"
|
|
);
|
|
|
|
Ok(())
|
|
}
|
|
|
|
// Verifies that a standard tool call (shell_command) exceeding the model formatting
|
|
// limits is truncated before being sent back to the model.
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn tool_call_output_exceeds_limit_truncated_chars_limit() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
|
|
// Use a model that exposes the shell_command tool.
|
|
let mut builder = test_codex().with_model("gpt-5.1");
|
|
|
|
let fixture = builder.build(&server).await?;
|
|
|
|
let call_id = "shell-too-large";
|
|
let command = if cfg!(windows) {
|
|
"for ($i=1; $i -le 100000; $i++) { Write-Output $i }"
|
|
} else {
|
|
"seq 1 100000"
|
|
};
|
|
let args = serde_json::json!({
|
|
"command": command,
|
|
"timeout_ms": 5_000,
|
|
});
|
|
|
|
// First response: model tells us to run the tool; second: complete the turn.
|
|
mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_response_created("resp-1"),
|
|
responses::ev_function_call(call_id, "shell_command", &serde_json::to_string(&args)?),
|
|
responses::ev_completed("resp-1"),
|
|
]),
|
|
)
|
|
.await;
|
|
let mock2 = mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_assistant_message("msg-1", "done"),
|
|
responses::ev_completed("resp-2"),
|
|
]),
|
|
)
|
|
.await;
|
|
|
|
fixture
|
|
.submit_turn_with_policy("trigger big shell output", SandboxPolicy::DangerFullAccess)
|
|
.await?;
|
|
|
|
// Inspect what we sent back to the model; it should contain a truncated
|
|
// function_call_output for the shell call.
|
|
let output = mock2
|
|
.single_request()
|
|
.function_call_output_text(call_id)
|
|
.context("function_call_output present for shell call")?;
|
|
let output = output.replace("\r\n", "\n");
|
|
|
|
// Expect plain text (not JSON) containing the entire shell output.
|
|
assert!(
|
|
serde_json::from_str::<Value>(&output).is_err(),
|
|
"expected truncated shell output to be plain text"
|
|
);
|
|
|
|
let truncated_pattern = r#"(?s)^Exit code: 0\nWall time: [0-9]+(?:\.[0-9]+)? seconds\nTotal output lines: 100000\nOutput:\n.*?…\d+ chars truncated….*$"#;
|
|
|
|
assert_regex_match(truncated_pattern, &output);
|
|
|
|
let len = output.len();
|
|
assert!(
|
|
(9_900..=10_100).contains(&len),
|
|
"expected ~10k chars after truncation, got {len}"
|
|
);
|
|
|
|
Ok(())
|
|
}
|
|
|
|
// Verifies that a standard tool call (shell_command) exceeding the model formatting
|
|
// limits is truncated before being sent back to the model.
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn tool_call_output_exceeds_limit_truncated_for_model() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
|
|
// Use a model that exposes the shell_command tool.
|
|
let mut builder = test_codex().with_model("gpt-5.1-codex");
|
|
let fixture = builder.build(&server).await?;
|
|
|
|
let call_id = "shell-too-large";
|
|
let command = if cfg!(windows) {
|
|
"for ($i=1; $i -le 100000; $i++) { Write-Output $i }"
|
|
} else {
|
|
"seq 1 100000"
|
|
};
|
|
let args = serde_json::json!({
|
|
"command": command,
|
|
"timeout_ms": 5_000,
|
|
});
|
|
|
|
// First response: model tells us to run the tool; second: complete the turn.
|
|
mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_response_created("resp-1"),
|
|
responses::ev_function_call(call_id, "shell_command", &serde_json::to_string(&args)?),
|
|
responses::ev_completed("resp-1"),
|
|
]),
|
|
)
|
|
.await;
|
|
let mock2 = mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_assistant_message("msg-1", "done"),
|
|
responses::ev_completed("resp-2"),
|
|
]),
|
|
)
|
|
.await;
|
|
|
|
fixture
|
|
.submit_turn_with_policy("trigger big shell output", SandboxPolicy::DangerFullAccess)
|
|
.await?;
|
|
|
|
// Inspect what we sent back to the model; it should contain a truncated
|
|
// function_call_output for the shell call.
|
|
let output = mock2
|
|
.single_request()
|
|
.function_call_output_text(call_id)
|
|
.context("function_call_output present for shell call")?;
|
|
let output = output.replace("\r\n", "\n");
|
|
|
|
// Expect plain text (not JSON) containing the entire shell output.
|
|
assert!(
|
|
serde_json::from_str::<Value>(&output).is_err(),
|
|
"expected truncated shell output to be plain text"
|
|
);
|
|
let truncated_pattern = r#"(?s)^Exit code: 0
|
|
Wall time: [0-9]+(?:\.[0-9]+)? seconds
|
|
Total output lines: 100000
|
|
Output:
|
|
1
|
|
2
|
|
3
|
|
4
|
|
5
|
|
6
|
|
.*…137224 tokens truncated.*
|
|
99999
|
|
100000
|
|
$"#;
|
|
assert_regex_match(truncated_pattern, &output);
|
|
|
|
Ok(())
|
|
}
|
|
|
|
// Ensures shell_command outputs that exceed the line limit are truncated only once.
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn tool_call_output_truncated_only_once() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
|
|
let mut builder = test_codex().with_model("gpt-5.1-codex");
|
|
let fixture = builder.build(&server).await?;
|
|
let call_id = "shell-single-truncation";
|
|
let command = if cfg!(windows) {
|
|
"for ($i=1; $i -le 10000; $i++) { Write-Output $i }"
|
|
} else {
|
|
"seq 1 10000"
|
|
};
|
|
let args = serde_json::json!({
|
|
"command": command,
|
|
"timeout_ms": 5_000,
|
|
});
|
|
|
|
mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_response_created("resp-1"),
|
|
responses::ev_function_call(call_id, "shell_command", &serde_json::to_string(&args)?),
|
|
responses::ev_completed("resp-1"),
|
|
]),
|
|
)
|
|
.await;
|
|
let mock2 = mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_assistant_message("msg-1", "done"),
|
|
responses::ev_completed("resp-2"),
|
|
]),
|
|
)
|
|
.await;
|
|
|
|
fixture
|
|
.submit_turn_with_policy("trigger big shell output", SandboxPolicy::DangerFullAccess)
|
|
.await?;
|
|
|
|
let output = mock2
|
|
.single_request()
|
|
.function_call_output_text(call_id)
|
|
.context("function_call_output present for shell call")?;
|
|
|
|
let truncation_markers = output.matches("tokens truncated").count();
|
|
|
|
assert_eq!(
|
|
truncation_markers, 1,
|
|
"shell output should carry only one truncation marker: {output}"
|
|
);
|
|
|
|
Ok(())
|
|
}
|
|
|
|
// Verifies that an MCP tool call result exceeding the model formatting limits
|
|
// is truncated before being sent back to the model.
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 1)]
|
|
async fn mcp_tool_call_output_exceeds_limit_truncated_for_model() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
|
|
let call_id = "rmcp-truncated";
|
|
let server_name = "rmcp";
|
|
let tool_name = format!("mcp__{server_name}__echo");
|
|
|
|
// Build a very large message to exceed 10KiB once serialized.
|
|
let large_msg = "long-message-with-newlines-".repeat(6000);
|
|
let args_json = serde_json::json!({ "message": large_msg });
|
|
|
|
mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_response_created("resp-1"),
|
|
responses::ev_function_call(call_id, &tool_name, &args_json.to_string()),
|
|
responses::ev_completed("resp-1"),
|
|
]),
|
|
)
|
|
.await;
|
|
let mock2 = mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_assistant_message("msg-1", "rmcp echo tool completed."),
|
|
responses::ev_completed("resp-2"),
|
|
]),
|
|
)
|
|
.await;
|
|
|
|
// Compile the rmcp stdio test server and configure it.
|
|
let rmcp_test_server_bin = stdio_server_bin()?;
|
|
|
|
let mut builder = test_codex().with_config(move |config| {
|
|
let mut servers = config.mcp_servers.get().clone();
|
|
servers.insert(
|
|
server_name.to_string(),
|
|
codex_core::config::types::McpServerConfig {
|
|
transport: codex_core::config::types::McpServerTransportConfig::Stdio {
|
|
command: rmcp_test_server_bin,
|
|
args: Vec::new(),
|
|
env: None,
|
|
env_vars: Vec::new(),
|
|
cwd: None,
|
|
},
|
|
enabled: true,
|
|
disabled_reason: None,
|
|
startup_timeout_sec: Some(std::time::Duration::from_secs(10)),
|
|
tool_timeout_sec: None,
|
|
enabled_tools: None,
|
|
disabled_tools: None,
|
|
scopes: None,
|
|
},
|
|
);
|
|
config
|
|
.mcp_servers
|
|
.set(servers)
|
|
.expect("test mcp servers should accept any configuration");
|
|
config.tool_output_token_limit = Some(500);
|
|
});
|
|
let fixture = builder.build(&server).await?;
|
|
|
|
fixture
|
|
.submit_turn_with_policy(
|
|
"call the rmcp echo tool with a very large message",
|
|
SandboxPolicy::ReadOnly,
|
|
)
|
|
.await?;
|
|
|
|
// The MCP tool call output is converted to a function_call_output for the model.
|
|
let output = mock2
|
|
.single_request()
|
|
.function_call_output_text(call_id)
|
|
.context("function_call_output present for rmcp call")?;
|
|
|
|
assert!(
|
|
!output.contains("Total output lines:"),
|
|
"MCP output should not include line-based truncation header: {output}"
|
|
);
|
|
|
|
let truncated_pattern = r#"(?s)^\{"echo":\s*"ECHOING: long-message-with-newlines-.*tokens truncated.*long-message-with-newlines-.*$"#;
|
|
assert_regex_match(truncated_pattern, &output);
|
|
assert!(output.len() < 2500, "{}", output.len());
|
|
|
|
Ok(())
|
|
}
|
|
|
|
// Verifies that an MCP image tool output is serialized as content_items array with
|
|
// the image preserved and no truncation summary appended (since there are no text items).
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 1)]
|
|
async fn mcp_image_output_preserves_image_and_no_text_summary() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
|
|
let call_id = "rmcp-image-no-trunc";
|
|
let server_name = "rmcp";
|
|
let tool_name = format!("mcp__{server_name}__image");
|
|
|
|
mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
ev_response_created("resp-1"),
|
|
ev_function_call(call_id, &tool_name, "{}"),
|
|
ev_completed("resp-1"),
|
|
]),
|
|
)
|
|
.await;
|
|
let final_mock = mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
ev_assistant_message("msg-1", "done"),
|
|
ev_completed("resp-2"),
|
|
]),
|
|
)
|
|
.await;
|
|
|
|
// Build the stdio rmcp server and pass a tiny PNG via data URL so it can construct ImageContent.
|
|
let rmcp_test_server_bin = stdio_server_bin()?;
|
|
|
|
// 1x1 PNG data URL
|
|
let openai_png = "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mP8/x8AAwMB/ee9bQAAAABJRU5ErkJggg==";
|
|
|
|
let mut builder = test_codex().with_config(move |config| {
|
|
let mut servers = config.mcp_servers.get().clone();
|
|
servers.insert(
|
|
server_name.to_string(),
|
|
McpServerConfig {
|
|
transport: McpServerTransportConfig::Stdio {
|
|
command: rmcp_test_server_bin,
|
|
args: Vec::new(),
|
|
env: Some(HashMap::from([(
|
|
"MCP_TEST_IMAGE_DATA_URL".to_string(),
|
|
openai_png.to_string(),
|
|
)])),
|
|
env_vars: Vec::new(),
|
|
cwd: None,
|
|
},
|
|
enabled: true,
|
|
disabled_reason: None,
|
|
startup_timeout_sec: Some(Duration::from_secs(10)),
|
|
tool_timeout_sec: None,
|
|
enabled_tools: None,
|
|
disabled_tools: None,
|
|
scopes: None,
|
|
},
|
|
);
|
|
config
|
|
.mcp_servers
|
|
.set(servers)
|
|
.expect("test mcp servers should accept any configuration");
|
|
});
|
|
let fixture = builder.build(&server).await?;
|
|
let session_model = fixture.session_configured.model.clone();
|
|
|
|
fixture
|
|
.codex
|
|
.submit(Op::UserTurn {
|
|
items: vec![UserInput::Text {
|
|
text: "call the rmcp image tool".into(),
|
|
text_elements: Vec::new(),
|
|
}],
|
|
final_output_json_schema: None,
|
|
cwd: fixture.cwd.path().to_path_buf(),
|
|
approval_policy: AskForApproval::Never,
|
|
sandbox_policy: SandboxPolicy::ReadOnly,
|
|
model: session_model,
|
|
effort: None,
|
|
summary: ReasoningSummary::Auto,
|
|
collaboration_mode: None,
|
|
personality: None,
|
|
})
|
|
.await?;
|
|
|
|
// Wait for completion to ensure the outbound request is captured.
|
|
wait_for_event(&fixture.codex, |ev| matches!(ev, EventMsg::TurnComplete(_))).await;
|
|
let output_item = final_mock.single_request().function_call_output(call_id);
|
|
// Expect exactly one array element: the image item; and no trailing summary text.
|
|
let output = output_item.get("output").expect("output");
|
|
assert!(output.is_array(), "expected array output");
|
|
let arr = output.as_array().unwrap();
|
|
assert_eq!(arr.len(), 1, "no truncation summary should be appended");
|
|
assert_eq!(
|
|
arr[0],
|
|
json!({"type": "input_image", "image_url": openai_png})
|
|
);
|
|
|
|
Ok(())
|
|
}
|
|
|
|
// Token-based policy should report token counts even when truncation is byte-estimated.
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn token_policy_marker_reports_tokens() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
let mut builder = test_codex()
|
|
.with_model("gpt-5.1-codex")
|
|
.with_config(|config| {
|
|
config.tool_output_token_limit = Some(50); // small budget to force truncation
|
|
});
|
|
let fixture = builder.build(&server).await?;
|
|
|
|
let call_id = "shell-token-marker";
|
|
let args = json!({
|
|
"command": "seq 1 150",
|
|
"timeout_ms": 5_000,
|
|
});
|
|
|
|
mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
ev_response_created("resp-1"),
|
|
ev_function_call(call_id, "shell_command", &serde_json::to_string(&args)?),
|
|
ev_completed("resp-1"),
|
|
]),
|
|
)
|
|
.await;
|
|
let done_mock = mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
ev_assistant_message("msg-1", "done"),
|
|
ev_completed("resp-2"),
|
|
]),
|
|
)
|
|
.await;
|
|
|
|
fixture
|
|
.submit_turn_with_policy("run the shell tool", SandboxPolicy::DangerFullAccess)
|
|
.await?;
|
|
|
|
let output = done_mock
|
|
.single_request()
|
|
.function_call_output_text(call_id)
|
|
.context("shell output present")?;
|
|
|
|
let pattern = r"(?s)^Exit code: 0\nWall time: [0-9]+(?:\.[0-9]+)? seconds\nTotal output lines: 150\nOutput:\n1\n2\n3\n4\n5\n6\n7\n8\n9\n10\n11\n12\n13\n14\n15\n16\n17\n18\n19.*tokens truncated.*129\n130\n131\n132\n133\n134\n135\n136\n137\n138\n139\n140\n141\n142\n143\n144\n145\n146\n147\n148\n149\n150\n$";
|
|
|
|
assert_regex_match(pattern, &output);
|
|
|
|
Ok(())
|
|
}
|
|
|
|
// Byte-based policy should report bytes removed.
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn byte_policy_marker_reports_bytes() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
let mut builder = test_codex().with_model("gpt-5.1").with_config(|config| {
|
|
config.tool_output_token_limit = Some(50); // ~200 byte cap
|
|
});
|
|
let fixture = builder.build(&server).await?;
|
|
|
|
let call_id = "shell-byte-marker";
|
|
let args = json!({
|
|
"command": "seq 1 150",
|
|
"timeout_ms": 5_000,
|
|
});
|
|
|
|
mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
ev_response_created("resp-1"),
|
|
ev_function_call(call_id, "shell_command", &serde_json::to_string(&args)?),
|
|
ev_completed("resp-1"),
|
|
]),
|
|
)
|
|
.await;
|
|
let done_mock = mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
ev_assistant_message("msg-1", "done"),
|
|
ev_completed("resp-2"),
|
|
]),
|
|
)
|
|
.await;
|
|
|
|
fixture
|
|
.submit_turn_with_policy("run the shell tool", SandboxPolicy::DangerFullAccess)
|
|
.await?;
|
|
|
|
let output = done_mock
|
|
.single_request()
|
|
.function_call_output_text(call_id)
|
|
.context("shell output present")?;
|
|
|
|
let pattern = r"(?s)^Exit code: 0\nWall time: [0-9]+(?:\.[0-9]+)? seconds\nTotal output lines: 150\nOutput:\n1\n2\n3\n4\n5\n6\n7\n8\n9\n10\n11\n12\n13\n14\n15\n16\n17\n18\n19.*chars truncated.*129\n130\n131\n132\n133\n134\n135\n136\n137\n138\n139\n140\n141\n142\n143\n144\n145\n146\n147\n148\n149\n150\n$";
|
|
|
|
assert_regex_match(pattern, &output);
|
|
|
|
Ok(())
|
|
}
|
|
|
|
// shell_command output should remain intact when the config opts into a large token budget.
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn shell_command_output_not_truncated_with_custom_limit() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
let mut builder = test_codex()
|
|
.with_model("gpt-5.1-codex")
|
|
.with_config(|config| {
|
|
config.tool_output_token_limit = Some(50_000); // ample budget
|
|
});
|
|
let fixture = builder.build(&server).await?;
|
|
|
|
let call_id = "shell-no-trunc";
|
|
let args = json!({
|
|
"command": "seq 1 1000",
|
|
"timeout_ms": 5_000,
|
|
});
|
|
let expected_body: String = (1..=1000).map(|i| format!("{i}\n")).collect();
|
|
|
|
mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
ev_response_created("resp-1"),
|
|
ev_function_call(call_id, "shell_command", &serde_json::to_string(&args)?),
|
|
ev_completed("resp-1"),
|
|
]),
|
|
)
|
|
.await;
|
|
let done_mock = mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
ev_assistant_message("msg-1", "done"),
|
|
ev_completed("resp-2"),
|
|
]),
|
|
)
|
|
.await;
|
|
|
|
fixture
|
|
.submit_turn_with_policy(
|
|
"run big output without truncation",
|
|
SandboxPolicy::DangerFullAccess,
|
|
)
|
|
.await?;
|
|
|
|
let output = done_mock
|
|
.single_request()
|
|
.function_call_output_text(call_id)
|
|
.context("shell output present")?;
|
|
|
|
assert!(
|
|
output.ends_with(&expected_body),
|
|
"expected entire shell output when budget increased: {output}"
|
|
);
|
|
assert!(
|
|
!output.contains("truncated"),
|
|
"output should remain untruncated with ample budget"
|
|
);
|
|
|
|
Ok(())
|
|
}
|
|
|
|
// MCP server output should also remain intact when the config increases the token limit.
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 1)]
|
|
async fn mcp_tool_call_output_not_truncated_with_custom_limit() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
|
|
let call_id = "rmcp-untruncated";
|
|
let server_name = "rmcp";
|
|
let tool_name = format!("mcp__{server_name}__echo");
|
|
let large_msg = "a".repeat(80_000);
|
|
let args_json = serde_json::json!({ "message": large_msg });
|
|
|
|
mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_response_created("resp-1"),
|
|
responses::ev_function_call(call_id, &tool_name, &args_json.to_string()),
|
|
responses::ev_completed("resp-1"),
|
|
]),
|
|
)
|
|
.await;
|
|
let mock2 = mount_sse_once(
|
|
&server,
|
|
sse(vec![
|
|
responses::ev_assistant_message("msg-1", "rmcp echo tool completed."),
|
|
responses::ev_completed("resp-2"),
|
|
]),
|
|
)
|
|
.await;
|
|
|
|
let rmcp_test_server_bin = stdio_server_bin()?;
|
|
|
|
let mut builder = test_codex().with_config(move |config| {
|
|
config.tool_output_token_limit = Some(50_000);
|
|
let mut servers = config.mcp_servers.get().clone();
|
|
servers.insert(
|
|
server_name.to_string(),
|
|
codex_core::config::types::McpServerConfig {
|
|
transport: codex_core::config::types::McpServerTransportConfig::Stdio {
|
|
command: rmcp_test_server_bin,
|
|
args: Vec::new(),
|
|
env: None,
|
|
env_vars: Vec::new(),
|
|
cwd: None,
|
|
},
|
|
enabled: true,
|
|
disabled_reason: None,
|
|
startup_timeout_sec: Some(std::time::Duration::from_secs(10)),
|
|
tool_timeout_sec: None,
|
|
enabled_tools: None,
|
|
disabled_tools: None,
|
|
scopes: None,
|
|
},
|
|
);
|
|
config
|
|
.mcp_servers
|
|
.set(servers)
|
|
.expect("test mcp servers should accept any configuration");
|
|
});
|
|
let fixture = builder.build(&server).await?;
|
|
|
|
fixture
|
|
.submit_turn_with_policy(
|
|
"call the rmcp echo tool with a very large message",
|
|
SandboxPolicy::ReadOnly,
|
|
)
|
|
.await?;
|
|
|
|
let output = mock2
|
|
.single_request()
|
|
.function_call_output_text(call_id)
|
|
.context("function_call_output present for rmcp call")?;
|
|
|
|
let parsed: Value = serde_json::from_str(&output)?;
|
|
assert_eq!(
|
|
output.len(),
|
|
80031,
|
|
"parsed MCP output should retain its serialized length"
|
|
);
|
|
let expected_echo = format!("ECHOING: {large_msg}");
|
|
let echo_str = parsed["echo"]
|
|
.as_str()
|
|
.context("echo field should be a string in rmcp echo output")?;
|
|
assert_eq!(
|
|
echo_str.len(),
|
|
expected_echo.len(),
|
|
"echo length should match"
|
|
);
|
|
assert_eq!(echo_str, expected_echo);
|
|
assert!(
|
|
!output.contains("truncated"),
|
|
"output should not include truncation markers when limit is raised: {output}"
|
|
);
|
|
|
|
Ok(())
|
|
}
|