feat(stdio): add auto-start connections and orphan process cleanup

[vibegui]

Summary by cubic

Adds auto-start for STDIO connections on startup and cleans up orphaned processes. Adds a 5-minute cooldown after spawn failures, improves event bus delivery reliability, and enables gateway subscriptions.

• New Features

  • Auto-start via AUTO_START_CONNECTIONS (comma-separated titles). Waits for resetStdioConnectionPool, adds a short delay, then spawns using dangerouslyCreateSuperUserMCPProxy.
  • Orphan cleanup on startup: kills mesh-bridge, pilot, and any process on port 9999 from previous runs.
  • Event bus subscriptions via gateway: support subscriberId in EventSubscribe and pass MESH_CONNECTION_ID to STDIO servers.

• Bug Fixes

  • Suppress “Method not found” (-32601) errors from listPrompts to reduce auto-start log noise.
  • Event bus worker: prevent missed notifications on concurrent processNow calls; add helpful logs.
  • Add 5-minute cooldown after STDIO spawn failures to avoid retry loops; return a clear “Spawn failed” error.

^{Written for commit d6da145. Summary will update on new commits.}

[github-actions]

🧪 Benchmark

Should we run the MCP Gateway benchmark for this PR?

React with 👍 to run the benchmark.

Reaction	Action
👍	Run quick benchmark (10 & 128 tools)

Benchmark will run on the next push after you react.

[github-actions]

Release Options

Should a new version be published when this PR is merged?

React with an emoji to vote on the release type:

Reaction	Type	Next Version
👍	Prerelease	1.1.2-alpha.1
🎉	Patch	1.1.2
❤️	Minor	1.2.0
🚀	Major	2.0.0

Current version: 1.1.1

Deployment

• Deploy to production (triggers ArgoCD sync after Docker image is published)

[cubic-dev-ai]

2 issues found across 2 files

Prompt for AI agents (all issues)

Check if these issues are valid — if so, understand the root cause of each and fix them.


<file name="apps/mesh/src/api/app.ts">

<violation number="1" location="apps/mesh/src/api/app.ts:191">
P2: The `.catch()` on `resetStdioConnectionPool()` converts rejected promises to resolved ones, so `await poolResetPromise` will succeed even if pool reset failed. This defeats the stated purpose of waiting for completion before auto-start. Consider re-throwing after logging, or handling the error case explicitly in the auto-start IIFE.</violation>
</file>

<file name="apps/mesh/src/stdio/stable-transport.ts">

<violation number="1" location="apps/mesh/src/stdio/stable-transport.ts:374">
P1: Killing any process on port 9999 is dangerous. Unlike the `pkill` patterns above that match specific command lines, this kills ANY process on port 9999 without verification. Consider checking the process command line before killing, or using a more specific approach like `lsof -t -i:9999 | xargs -I{} sh -c &#39;ps -p {} -o args= | grep -q mesh-bridge &amp;&amp; kill -9 {}&#39;`.</violation>
</file>

_{Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review.}

[cubic-dev-ai]

P1: Killing any process on port 9999 is dangerous. Unlike the pkill patterns above that match specific command lines, this kills ANY process on port 9999 without verification. Consider checking the process command line before killing, or using a more specific approach like lsof -t -i:9999 | xargs -I{} sh -c 'ps -p {} -o args= | grep -q mesh-bridge && kill -9 {}'.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At apps/mesh/src/stdio/stable-transport.ts, line 374:

<comment>Killing any process on port 9999 is dangerous. Unlike the `pkill` patterns above that match specific command lines, this kills ANY process on port 9999 without verification. Consider checking the process command line before killing, or using a more specific approach like `lsof -t -i:9999 | xargs -I{} sh -c &#39;ps -p {} -o args= | grep -q mesh-bridge &amp;&amp; kill -9 {}&#39;`.</comment>

<file context>
@@ -343,6 +343,51 @@ async function forceCloseAllStdioConnections(): Promise&lt;void&gt; {
+
+  // Also kill anything listening on port 9999 (Bridge WebSocket)
+  try {
+    const { stdout } = await execAsync(`lsof -t -i:9999 2&gt;/dev/null || true`);
+    const pids = stdout.trim().split(&quot;\n&quot;).filter(Boolean);
+    for (const pid of pids) {
</file context>

✅ Addressed in c662c2a

[cubic-dev-ai]

P2: The .catch() on resetStdioConnectionPool() converts rejected promises to resolved ones, so await poolResetPromise will succeed even if pool reset failed. This defeats the stated purpose of waiting for completion before auto-start. Consider re-throwing after logging, or handling the error case explicitly in the auto-start IIFE.

Prompt for AI agents

Check if this issue is valid — if so, understand the root cause and fix it. At apps/mesh/src/api/app.ts, line 191:

<comment>The `.catch()` on `resetStdioConnectionPool()` converts rejected promises to resolved ones, so `await poolResetPromise` will succeed even if pool reset failed. This defeats the stated purpose of waiting for completion before auto-start. Consider re-throwing after logging, or handling the error case explicitly in the auto-start IIFE.</comment>

<file context>
@@ -129,7 +187,8 @@ export function createApp(options: CreateAppOptions = {}) {
   // Old processes have stale credentials, need fresh spawn with new tokens
-  resetStdioConnectionPool().catch((err) =&gt; {
+  // IMPORTANT: Track this promise so autoStart waits for it to complete
+  const poolResetPromise = resetStdioConnectionPool().catch((err) =&gt; {
     console.error(&quot;[StableStdio] Error resetting pool:&quot;, err);
   });
</file context>

✅ Addressed in c662c2a