mirror of
https://github.com/openai/codex.git
synced 2026-11-06 00:30:29 +00:00
## What changed - Add an optional `policy_template` field to auto-review model messages. - Build Guardian instructions from the catalog template when present, while preserving the existing precedence for managed and catalog policy content. - Fall back independently to the bundled template and bundled policy, and preserve explicit empty catalog values. ## Testing - Cover missing and empty template deserialization, policy precedence, bundled fallbacks, and Guardian prewarm instructions. GitOrigin-RevId: 93116dadbc9bf40527882844a428caed4159fff6
394 lines
15 KiB
Rust
394 lines
15 KiB
Rust
#![cfg(not(target_os = "windows"))]
|
|
|
|
use anyhow::Result;
|
|
use codex_core::config::Constrained;
|
|
use codex_core::sandboxing::SandboxPermissions;
|
|
use codex_protocol::config_types::ApprovalsReviewer;
|
|
use codex_protocol::openai_models::AutoReviewMessages;
|
|
use codex_protocol::openai_models::ModelsResponse;
|
|
use codex_protocol::protocol::AskForApproval;
|
|
use codex_protocol::protocol::EventMsg;
|
|
use codex_protocol::protocol::Op;
|
|
use codex_protocol::protocol::SandboxPolicy;
|
|
use codex_protocol::user_input::UserInput;
|
|
use core_test_support::fs_wait;
|
|
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_sequence;
|
|
use core_test_support::responses::sse;
|
|
use core_test_support::responses::start_mock_server;
|
|
use core_test_support::responses::start_websocket_server;
|
|
use core_test_support::skip_if_no_network;
|
|
use core_test_support::skip_if_sandbox;
|
|
use core_test_support::skip_if_wine_exec;
|
|
use core_test_support::test_codex::local_selections;
|
|
use core_test_support::test_codex::test_codex;
|
|
use core_test_support::wait_for_event;
|
|
use pretty_assertions::assert_eq;
|
|
use serde_json::Value;
|
|
use serde_json::json;
|
|
use std::fs;
|
|
use std::os::unix::fs::PermissionsExt;
|
|
use std::time::Duration;
|
|
use tempfile::TempDir;
|
|
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn guardian_session_prewarms_and_is_reused_for_first_review() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
|
|
let catalog_template = "Catalog-provided Guardian template:\n{{ tenant_policy_config }}";
|
|
let mut review_model = codex_models_manager::bundled_models_response()?
|
|
.models
|
|
.into_iter()
|
|
.find(|model| model.slug == "codex-auto-review")
|
|
.expect("bundled auto-review model");
|
|
let model_messages = review_model
|
|
.model_messages
|
|
.as_mut()
|
|
.expect("auto-review model messages");
|
|
model_messages.auto_review = Some(AutoReviewMessages {
|
|
policy: None,
|
|
policy_template: Some(catalog_template.to_string()),
|
|
});
|
|
|
|
let tool_args = json!({
|
|
"cmd": "true",
|
|
"sandbox_permissions": SandboxPermissions::RequireEscalated,
|
|
"justification": "Exercise Guardian approval routing.",
|
|
})
|
|
.to_string();
|
|
let server = start_websocket_server(vec![
|
|
vec![vec![ev_response_created("warm-1"), ev_completed("warm-1")]],
|
|
vec![vec![ev_response_created("warm-2"), ev_completed("warm-2")]],
|
|
vec![vec![
|
|
ev_response_created("approval-request"),
|
|
ev_function_call("approval-call", "exec_command", &tool_args),
|
|
ev_completed("approval-request"),
|
|
]],
|
|
vec![vec![
|
|
ev_response_created("guardian-review"),
|
|
ev_completed("guardian-review"),
|
|
]],
|
|
])
|
|
.await;
|
|
let mut builder = test_codex().with_config(move |config| {
|
|
config.model_catalog = Some(ModelsResponse {
|
|
models: vec![review_model],
|
|
});
|
|
config.permissions.approval_policy = Constrained::allow_any(AskForApproval::OnRequest);
|
|
config.approvals_reviewer = ApprovalsReviewer::AutoReview;
|
|
});
|
|
|
|
let test = builder.build_with_websocket_server(&server).await?;
|
|
let (first, second) = tokio::time::timeout(Duration::from_secs(5), async {
|
|
tokio::join!(
|
|
server.wait_for_request(/*connection_index*/ 0, /*request_index*/ 0),
|
|
server.wait_for_request(/*connection_index*/ 1, /*request_index*/ 0)
|
|
)
|
|
})
|
|
.await?;
|
|
let prewarm_requests = [first.body_json(), second.body_json()];
|
|
let guardian_prewarm = prewarm_requests
|
|
.iter()
|
|
.find(|request| {
|
|
request["client_metadata"]["x-openai-subagent"].as_str() == Some("guardian")
|
|
})
|
|
.expect("guardian startup prewarm request");
|
|
assert_eq!(guardian_prewarm["generate"].as_bool(), Some(false));
|
|
let guardian_instructions = guardian_prewarm["instructions"]
|
|
.as_str()
|
|
.expect("guardian instructions");
|
|
assert!(guardian_instructions.contains("Catalog-provided Guardian template:"));
|
|
assert!(guardian_instructions.contains("- Organization: default generic tenant."));
|
|
assert!(!guardian_instructions.contains("{{ tenant_policy_config }}"));
|
|
assert!(guardian_instructions.contains("final message must be strict JSON"));
|
|
let guardian_thread_id = guardian_prewarm["client_metadata"]["thread_id"]
|
|
.as_str()
|
|
.expect("guardian thread id");
|
|
|
|
test.codex
|
|
.submit(
|
|
vec![UserInput::Text {
|
|
text: "run a command that requires Guardian review".into(),
|
|
text_elements: Vec::new(),
|
|
}]
|
|
.into(),
|
|
)
|
|
.await?;
|
|
let guardian_review = tokio::time::timeout(
|
|
Duration::from_secs(5),
|
|
server.wait_for_request(/*connection_index*/ 3, /*request_index*/ 0),
|
|
)
|
|
.await?
|
|
.body_json();
|
|
assert_eq!(
|
|
guardian_review["client_metadata"]["x-openai-subagent"].as_str(),
|
|
Some("guardian")
|
|
);
|
|
assert_eq!(
|
|
guardian_review["client_metadata"]["thread_id"].as_str(),
|
|
Some(guardian_thread_id)
|
|
);
|
|
assert_eq!(guardian_review.get("generate"), None);
|
|
|
|
test.codex.shutdown_and_wait().await?;
|
|
server.shutdown().await;
|
|
Ok(())
|
|
}
|
|
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn guardian_denial_rejects_tool_call_with_rationale() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
skip_if_sandbox!(Ok(()));
|
|
skip_if_wine_exec!(
|
|
Ok(()),
|
|
"Guardian approval actions require host-native paths"
|
|
);
|
|
|
|
let server = start_mock_server().await;
|
|
let approval_policy = AskForApproval::OnRequest;
|
|
let sandbox_policy = SandboxPolicy::WorkspaceWrite {
|
|
writable_roots: vec![],
|
|
network_access: false,
|
|
exclude_tmpdir_env_var: true,
|
|
exclude_slash_tmp: true,
|
|
};
|
|
let sandbox_policy_for_config = sandbox_policy.clone();
|
|
|
|
let mut builder = test_codex().with_config(move |config| {
|
|
config.permissions.approval_policy = Constrained::allow_any(approval_policy);
|
|
config
|
|
.set_legacy_sandbox_policy(sandbox_policy_for_config)
|
|
.expect("set sandbox policy");
|
|
});
|
|
let test = builder.build_with_auto_env(&server).await?;
|
|
|
|
let output_file = test.cwd.path().join("guardian-denied.txt");
|
|
let command = format!("printf should-not-run > {}", output_file.display());
|
|
let tool_args = json!({
|
|
"cmd": command,
|
|
"yield_time_ms": 1_000_u64,
|
|
"sandbox_permissions": SandboxPermissions::RequireEscalated,
|
|
"justification": "Exercise Guardian denial routing.",
|
|
});
|
|
let responses = mount_sse_sequence(
|
|
&server,
|
|
vec![
|
|
sse(vec![
|
|
ev_response_created("resp-parent-tool-denied"),
|
|
ev_function_call(
|
|
"exec-call-denied",
|
|
"exec_command",
|
|
&serde_json::to_string(&tool_args)?,
|
|
),
|
|
ev_completed("resp-parent-tool-denied"),
|
|
]),
|
|
sse(vec![
|
|
ev_response_created("resp-guardian-denied"),
|
|
ev_assistant_message(
|
|
"msg-guardian-denied",
|
|
&json!({
|
|
"risk_level": "high",
|
|
"user_authorization": "low",
|
|
"outcome": "deny",
|
|
"rationale": "The requested write has unacceptable test risk.",
|
|
})
|
|
.to_string(),
|
|
),
|
|
ev_completed("resp-guardian-denied"),
|
|
]),
|
|
sse(vec![
|
|
ev_response_created("resp-parent-after-denial"),
|
|
ev_assistant_message("msg-parent-after-denial", "denied"),
|
|
ev_completed("resp-parent-after-denial"),
|
|
]),
|
|
],
|
|
)
|
|
.await;
|
|
|
|
test.codex
|
|
.submit(Op::UserInput {
|
|
items: vec![UserInput::Text {
|
|
text: "run a command that Guardian should deny".into(),
|
|
text_elements: Vec::new(),
|
|
}],
|
|
final_output_json_schema: None,
|
|
responsesapi_client_metadata: None,
|
|
additional_context: Default::default(),
|
|
thread_settings: codex_protocol::protocol::ThreadSettingsOverrides {
|
|
approval_policy: Some(approval_policy),
|
|
approvals_reviewer: Some(ApprovalsReviewer::AutoReview),
|
|
sandbox_policy: Some(sandbox_policy),
|
|
..Default::default()
|
|
},
|
|
})
|
|
.await?;
|
|
wait_for_event(&test.codex, |event| {
|
|
matches!(event, EventMsg::TurnComplete(_))
|
|
})
|
|
.await;
|
|
|
|
let requests = responses.requests();
|
|
let guardian_request = requests
|
|
.iter()
|
|
.find(|request| request.body_contains_text("Exercise Guardian denial routing."))
|
|
.expect("expected Guardian review request");
|
|
assert!(guardian_request.body_contains_text(&command));
|
|
|
|
let tool_output = requests
|
|
.iter()
|
|
.find_map(|request| request.function_call_output_text("exec-call-denied"))
|
|
.expect("expected rejected tool output to be returned to the parent model");
|
|
assert!(
|
|
tool_output.contains("The requested write has unacceptable test risk."),
|
|
"Guardian rationale missing from rejected tool output: {tool_output}"
|
|
);
|
|
assert!(
|
|
!output_file.exists(),
|
|
"Guardian-denied command unexpectedly executed"
|
|
);
|
|
|
|
Ok(())
|
|
}
|
|
|
|
#[tokio::test(flavor = "multi_thread", worker_threads = 2)]
|
|
async fn guardian_review_session_does_not_inherit_legacy_notify() -> Result<()> {
|
|
skip_if_no_network!(Ok(()));
|
|
skip_if_sandbox!(Ok(()));
|
|
|
|
let server = start_mock_server().await;
|
|
let approval_policy = AskForApproval::OnRequest;
|
|
let sandbox_policy = SandboxPolicy::WorkspaceWrite {
|
|
writable_roots: vec![],
|
|
network_access: false,
|
|
exclude_tmpdir_env_var: true,
|
|
exclude_slash_tmp: true,
|
|
};
|
|
|
|
let notify_dir = TempDir::new()?;
|
|
let notify_script = notify_dir.path().join("notify.sh");
|
|
fs::write(
|
|
¬ify_script,
|
|
r#"#!/bin/bash
|
|
set -e
|
|
payload_path="$(dirname "${0}")/notify.jsonl"
|
|
printf '%s\n' "${@: -1}" >> "${payload_path}""#,
|
|
)?;
|
|
fs::set_permissions(¬ify_script, fs::Permissions::from_mode(0o755))?;
|
|
let notify_file = notify_dir.path().join("notify.jsonl");
|
|
let notify_script_str = notify_script.to_str().unwrap().to_string();
|
|
let sandbox_policy_for_config = sandbox_policy.clone();
|
|
|
|
let mut builder = test_codex().with_config(move |config| {
|
|
config.notify = Some(vec![notify_script_str]);
|
|
config.permissions.approval_policy = Constrained::allow_any(approval_policy);
|
|
config
|
|
.set_legacy_sandbox_policy(sandbox_policy_for_config)
|
|
.expect("set sandbox policy");
|
|
});
|
|
let test = builder.build(&server).await?;
|
|
|
|
let output_file = test.cwd.path().join("guardian-review-notify.txt");
|
|
let command = format!("printf guardian-approved > {}", output_file.display());
|
|
let tool_args = json!({
|
|
"cmd": command,
|
|
"yield_time_ms": 1_000_u64,
|
|
"sandbox_permissions": SandboxPermissions::RequireEscalated,
|
|
"justification": "Exercise Guardian approval routing.",
|
|
});
|
|
let responses = mount_sse_sequence(
|
|
&server,
|
|
vec![
|
|
sse(vec![
|
|
ev_response_created("resp-parent-tool"),
|
|
ev_function_call(
|
|
"exec-call",
|
|
"exec_command",
|
|
&serde_json::to_string(&tool_args)?,
|
|
),
|
|
ev_completed("resp-parent-tool"),
|
|
]),
|
|
sse(vec![
|
|
ev_response_created("resp-guardian-review"),
|
|
ev_assistant_message(
|
|
"msg-guardian-review",
|
|
&json!({
|
|
"risk_level": "low",
|
|
"user_authorization": "high",
|
|
"outcome": "allow",
|
|
"rationale": "The command writes a marker file in the workspace.",
|
|
})
|
|
.to_string(),
|
|
),
|
|
ev_completed("resp-guardian-review"),
|
|
]),
|
|
sse(vec![
|
|
ev_response_created("resp-parent-done"),
|
|
ev_assistant_message("msg-parent-done", "done"),
|
|
ev_completed("resp-parent-done"),
|
|
]),
|
|
],
|
|
)
|
|
.await;
|
|
|
|
test.codex
|
|
.submit(Op::UserInput {
|
|
items: vec![UserInput::Text {
|
|
text: "run a command that requires Guardian review".into(),
|
|
text_elements: Vec::new(),
|
|
}],
|
|
final_output_json_schema: None,
|
|
responsesapi_client_metadata: None,
|
|
additional_context: Default::default(),
|
|
thread_settings: codex_protocol::protocol::ThreadSettingsOverrides {
|
|
environments: Some(local_selections(test.config.cwd.clone())),
|
|
approval_policy: Some(approval_policy),
|
|
approvals_reviewer: Some(ApprovalsReviewer::AutoReview),
|
|
sandbox_policy: Some(sandbox_policy),
|
|
..Default::default()
|
|
},
|
|
})
|
|
.await?;
|
|
wait_for_event(&test.codex, |event| {
|
|
matches!(event, EventMsg::TurnComplete(_))
|
|
})
|
|
.await;
|
|
|
|
let guardian_request = responses
|
|
.requests()
|
|
.into_iter()
|
|
.find(|request| request.body_contains_text("Exercise Guardian approval routing."))
|
|
.expect("expected Guardian review request");
|
|
assert!(guardian_request.body_contains_text(&command));
|
|
|
|
fs_wait::wait_for_path_exists(¬ify_file, Duration::from_secs(5)).await?;
|
|
tokio::time::sleep(Duration::from_millis(100)).await;
|
|
let notify_payload_raw = tokio::fs::read_to_string(¬ify_file).await?;
|
|
let payloads: Vec<Value> = notify_payload_raw
|
|
.lines()
|
|
.map(serde_json::from_str::<Value>)
|
|
.collect::<std::result::Result<_, _>>()?;
|
|
|
|
assert_eq!(
|
|
payloads.len(),
|
|
1,
|
|
"unexpected notify payloads: {payloads:?}"
|
|
);
|
|
assert_eq!(
|
|
payloads[0]["input-messages"],
|
|
json!(["run a command that requires Guardian review"])
|
|
);
|
|
assert_eq!(payloads[0]["last-assistant-message"], json!("done"));
|
|
assert!(
|
|
!notify_payload_raw.contains(
|
|
"The following is the Codex agent history whose request action you are assessing."
|
|
),
|
|
"Guardian review transcript leaked into legacy notify payload: {notify_payload_raw}"
|
|
);
|
|
assert_eq!(fs::read_to_string(&output_file)?, "guardian-approved");
|
|
|
|
Ok(())
|
|
}
|