fix(app-server): thread history redaction for remote clients (#22178)

## Summary

Remote clients can still receive large `thread/resume` histories when
prior turns include MCP tool call payloads or image-generation results.
This adds a temporary response-only redaction path for the known remote
client names.

Longer term we will move towards fully paginated APIs backed by SQLite.

## Changes

- Redact MCP tool call payload-bearing fields in `thread/resume`
responses for `codex_chatgpt_android_remote` and
`codex_chatgpt_ios_remote`.
- Drop `imageGeneration` items from those `thread/resume` responses.
- Keep redaction out of persisted rollout files, `thread/read`,
`thread/turns/list`, live notifications, and token usage replay.
- Cover the behavior with app-server helper tests and a v2 resume
integration test that checks both remote clients plus a non-target
control client.

## Testing

- `cargo test -p codex-app-server thread_resume_redaction`
- `cargo test -p codex-app-server
thread_resume_redacts_payloads_for_chatgpt_remote_clients`
This commit is contained in:
Owen Lin
2026-05-11 11:45:25 -07:00
committed by GitHub
parent 90bd445e7f
commit 7bddb3083d
6 changed files with 417 additions and 2 deletions
@@ -15,6 +15,7 @@ use app_test_support::to_response;
use app_test_support::write_chatgpt_auth;
use chrono::Utc;
use codex_app_server_protocol::AskForApproval;
use codex_app_server_protocol::ClientInfo;
use codex_app_server_protocol::CommandExecutionApprovalDecision;
use codex_app_server_protocol::CommandExecutionRequestApprovalResponse;
use codex_app_server_protocol::FileChangeApprovalDecision;
@@ -51,10 +52,14 @@ use codex_config::types::AuthCredentialsStoreMode;
use codex_login::REFRESH_TOKEN_URL_OVERRIDE_ENV_VAR;
use codex_protocol::ThreadId;
use codex_protocol::config_types::Personality;
use codex_protocol::mcp::CallToolResult;
use codex_protocol::models::ContentItem;
use codex_protocol::models::ResponseItem;
use codex_protocol::protocol::AgentMessageEvent;
use codex_protocol::protocol::EventMsg;
use codex_protocol::protocol::ImageGenerationEndEvent;
use codex_protocol::protocol::McpInvocation;
use codex_protocol::protocol::McpToolCallEndEvent;
use codex_protocol::protocol::SessionMeta;
use codex_protocol::protocol::SessionMetaLine;
use codex_protocol::protocol::SessionSource as RolloutSessionSource;
@@ -77,6 +82,7 @@ use std::io::Write;
use std::path::Path;
use std::path::PathBuf;
use std::process::Command;
use std::time::Duration;
use tempfile::TempDir;
use tokio::time::timeout;
use uuid::Uuid;
@@ -376,6 +382,203 @@ async fn thread_resume_returns_rollout_history() -> Result<()> {
Ok(())
}
#[tokio::test]
async fn thread_resume_redacts_payloads_for_chatgpt_remote_clients() -> Result<()> {
for client_name in ["codex_chatgpt_android_remote", "codex_chatgpt_ios_remote"] {
let remote_thread = resume_redaction_fixture(Some(client_name)).await?;
let remote_turn = remote_thread
.turns
.first()
.expect("remote resume should include a turn");
let remote_mcp_item = remote_turn
.items
.iter()
.find(|item| matches!(item, ThreadItem::McpToolCall { .. }))
.expect("remote resume should include redacted MCP item");
let ThreadItem::McpToolCall {
arguments,
result,
error,
..
} = remote_mcp_item
else {
unreachable!("matched MCP item");
};
assert_eq!(arguments, &json!("[redacted]"));
let result = result.as_ref().expect("redacted MCP result");
assert_eq!(
result.content,
vec![json!({
"type": "text",
"text": "[redacted]",
})]
);
assert_eq!(result.structured_content, None);
assert_eq!(result.meta, None);
assert_eq!(error, &None);
assert!(
!remote_turn
.items
.iter()
.any(|item| matches!(item, ThreadItem::ImageGeneration { .. })),
"remote resume should drop image generation items for {client_name}"
);
}
let normal_thread = resume_redaction_fixture(Some("some_other_client")).await?;
let normal_turn = normal_thread
.turns
.first()
.expect("normal resume should include a turn");
let normal_mcp_item = normal_turn
.items
.iter()
.find(|item| matches!(item, ThreadItem::McpToolCall { .. }))
.expect("normal resume should include MCP item");
let ThreadItem::McpToolCall {
arguments, result, ..
} = normal_mcp_item
else {
unreachable!("matched MCP item");
};
assert_eq!(arguments, &json!({"secret":"argument"}));
let result = result.as_ref().expect("normal MCP result");
assert_eq!(
result.content,
vec![json!({
"type": "text",
"text": "secret result",
})]
);
assert_eq!(
result.structured_content,
Some(json!({"secret":"structured"}))
);
assert_eq!(result.meta, Some(json!({"secret":"meta"})));
assert!(
normal_turn.items.iter().any(|item| matches!(
item,
ThreadItem::ImageGeneration {
result,
revised_prompt,
..
} if result == "base64-image-result"
&& revised_prompt.as_deref() == Some("secret revised prompt")
)),
"normal resume should keep image generation items"
);
Ok(())
}
async fn resume_redaction_fixture(
client_name: Option<&str>,
) -> Result<codex_app_server_protocol::Thread> {
let server = create_mock_responses_server_repeating_assistant("Done").await;
let codex_home = TempDir::new()?;
create_config_toml(codex_home.path(), &server.uri())?;
let filename_ts = "2025-01-05T12-00-00";
let meta_rfc3339 = "2025-01-05T12:00:00Z";
let conversation_id = create_fake_rollout(
codex_home.path(),
filename_ts,
meta_rfc3339,
"Saved user message",
Some("mock_provider"),
/*git_info*/ None,
)?;
append_resume_redaction_history(
codex_home.path(),
filename_ts,
meta_rfc3339,
&conversation_id,
)?;
let mut mcp = McpProcess::new(codex_home.path()).await?;
if let Some(client_name) = client_name {
let _ = timeout(
DEFAULT_READ_TIMEOUT,
mcp.initialize_with_client_info(ClientInfo {
name: client_name.to_string(),
title: None,
version: "0.1.0".to_string(),
}),
)
.await??;
} else {
timeout(DEFAULT_READ_TIMEOUT, mcp.initialize()).await??;
}
let resume_id = mcp
.send_thread_resume_request(ThreadResumeParams {
thread_id: conversation_id,
..Default::default()
})
.await?;
let resume_resp: JSONRPCResponse = timeout(
DEFAULT_READ_TIMEOUT,
mcp.read_stream_until_response_message(RequestId::Integer(resume_id)),
)
.await??;
let ThreadResumeResponse { thread, .. } = to_response::<ThreadResumeResponse>(resume_resp)?;
Ok(thread)
}
fn append_resume_redaction_history(
codex_home: &Path,
filename_ts: &str,
meta_rfc3339: &str,
conversation_id: &str,
) -> Result<()> {
let rollout_file_path = rollout_path(codex_home, filename_ts, conversation_id);
let persisted_rollout = std::fs::read_to_string(&rollout_file_path)?;
let appended_rollout = [
EventMsg::McpToolCallEnd(McpToolCallEndEvent {
call_id: "mcp-1".to_string(),
invocation: McpInvocation {
server: "docs".to_string(),
tool: "lookup".to_string(),
arguments: Some(json!({"secret":"argument"})),
},
mcp_app_resource_uri: Some("ui://widget/lookup.html".to_string()),
duration: Duration::from_millis(8),
result: Ok(CallToolResult {
content: vec![json!({
"type": "text",
"text": "secret result",
})],
structured_content: Some(json!({"secret":"structured"})),
is_error: Some(false),
meta: Some(json!({"secret":"meta"})),
}),
}),
EventMsg::ImageGenerationEnd(ImageGenerationEndEvent {
call_id: "ig-1".to_string(),
status: "completed".to_string(),
revised_prompt: Some("secret revised prompt".to_string()),
result: "base64-image-result".to_string(),
saved_path: Some(test_absolute_path("/tmp/ig-1.png")),
}),
]
.into_iter()
.map(|payload| {
Ok(json!({
"timestamp": meta_rfc3339,
"type": "event_msg",
"payload": serde_json::to_value(payload)?,
})
.to_string())
})
.collect::<Result<Vec<_>>>()?
.join("\n");
std::fs::write(
&rollout_file_path,
format!("{persisted_rollout}{appended_rollout}\n"),
)?;
Ok(())
}
#[tokio::test]
async fn thread_resume_can_skip_turns_for_metadata_only_resume() -> Result<()> {
let server = create_mock_responses_server_repeating_assistant("Done").await;