fix: change codex/sandbox-state/update from a notification to a request (#8142)

Historically, `accept_elicitation_for_prompt_rule()` was flaky because
we were using a notification to update the sandbox followed by a `shell`
tool request that we expected to be subject to the new sandbox config,
but because [rmcp](https://crates.io/crates/rmcp) MCP servers delegate
each incoming message to a new Tokio task, messages are not guaranteed
to be processed in order, so sometimes the `shell` tool call would run
before the notification was processed.

Prior to this PR, we relied on a generous `sleep()` between the
notification and the request to reduce the change of the test flaking
out.

This PR implements a proper fix, which is to use a _request_ instead of
a notification for the sandbox update so that we can wait for the
response to the sandbox request before sending the request to the
`shell` tool call. Previously, `rmcp` did not support custom requests,
but I fixed that in
https://github.com/modelcontextprotocol/rust-sdk/pull/590, which made it
into the `0.12.0` release (see #8288).

This PR updates `shell-tool-mcp` to expect
`"codex/sandbox-state/update"` as a _request_ instead of a notification
and sends the appropriate ack. Note this behavior is tied to our custom
`codex/sandbox-state` capability, which Codex honors as an MCP client,
which is why `core/src/mcp_connection_manager.rs` had to be updated as
part of this PR, as well.

This PR also updates the docs at `shell-tool-mcp/README.md`.
This commit is contained in:
Michael Bolin
2025-12-18 15:32:01 -08:00
committed by GitHub
parent 358a5baba0
commit 46baedd7cb
7 changed files with 98 additions and 62 deletions

View File

@@ -5,7 +5,7 @@ use std::time::Duration;
use anyhow::Context as _;
use anyhow::Result;
use codex_core::MCP_SANDBOX_STATE_CAPABILITY;
use codex_core::MCP_SANDBOX_STATE_NOTIFICATION;
use codex_core::MCP_SANDBOX_STATE_METHOD;
use codex_core::SandboxState;
use codex_core::protocol::SandboxPolicy;
use codex_execpolicy::Policy;
@@ -15,6 +15,8 @@ use rmcp::ServerHandler;
use rmcp::ServiceExt;
use rmcp::handler::server::router::tool::ToolRouter;
use rmcp::handler::server::wrapper::Parameters;
use rmcp::model::CustomRequest;
use rmcp::model::CustomResult;
use rmcp::model::*;
use rmcp::schemars;
use rmcp::service::RequestContext;
@@ -23,8 +25,8 @@ use rmcp::tool;
use rmcp::tool_handler;
use rmcp::tool_router;
use rmcp::transport::stdio;
use serde_json::json;
use tokio::sync::RwLock;
use tracing::debug;
use crate::posix::escalate_server::EscalateServer;
use crate::posix::escalate_server::{self};
@@ -146,6 +148,13 @@ impl ExecTool {
}
}
#[derive(Default)]
pub struct CodexSandboxStateUpdateMethod;
impl rmcp::model::ConstString for CodexSandboxStateUpdateMethod {
const VALUE: &'static str = MCP_SANDBOX_STATE_METHOD;
}
#[tool_handler]
impl ServerHandler for ExecTool {
fn get_info(&self) -> ServerInfo {
@@ -181,29 +190,33 @@ impl ServerHandler for ExecTool {
Ok(self.get_info())
}
async fn on_custom_notification(
async fn on_custom_request(
&self,
notification: rmcp::model::CustomNotification,
_context: rmcp::service::NotificationContext<rmcp::RoleServer>,
) {
let rmcp::model::CustomNotification { method, params, .. } = notification;
if method == MCP_SANDBOX_STATE_NOTIFICATION
&& let Some(params) = params
{
match serde_json::from_value::<SandboxState>(params) {
Ok(sandbox_state) => {
debug!(
?sandbox_state.sandbox_policy,
"received sandbox state notification"
);
let mut state = self.sandbox_state.write().await;
*state = Some(sandbox_state);
}
Err(err) => {
tracing::warn!(?err, "failed to deserialize sandbox state notification");
}
}
request: CustomRequest,
_context: rmcp::service::RequestContext<rmcp::RoleServer>,
) -> Result<CustomResult, McpError> {
let CustomRequest { method, params, .. } = request;
if method != MCP_SANDBOX_STATE_METHOD {
return Err(McpError::method_not_found::<CodexSandboxStateUpdateMethod>());
}
let Some(params) = params else {
return Err(McpError::invalid_params(
"missing params for sandbox state request".to_string(),
None,
));
};
let Ok(sandbox_state) = serde_json::from_value::<SandboxState>(params.clone()) else {
return Err(McpError::invalid_params(
"failed to deserialize sandbox state".to_string(),
Some(params),
));
};
*self.sandbox_state.write().await = Some(sandbox_state);
Ok(CustomResult::new(json!({})))
}
}