[v2026.4.29] [OpenClaw] Is it stable?

[github-actions]

NO 🔥 — v2026.4.29

Six hours after release, the bug tracker looks like a 3am incident channel. Hard pass.

Stability Score: 34

Affected components

• discord
• gateway
• imessage
• runtime
• whatsapp

Evidence

Issue	Area	Type	Severity	Penalty	Reason
openclaw/openclaw#75283	gateway	kernel	5	-5	2026.4.29: Gateway repeatedly spawns runtime-deps pnpm installs and starves event loop
openclaw/openclaw#75330	gateway	kernel	5	-5	Gateway event loop blocked during agent prep: max delay 32s, utilization 100%, API requests hang
openclaw/openclaw#75430	runtime	other	1	-1	TUI startup hangs at 100% CPU due to bundled plugin reload loop (2026.4.29)
openclaw/openclaw#75394	whatsapp	channel-specific	1	-1	[Bug]: Failed to render WhatsApp QR - Cannot find package ‘qrcode’ in 2026.4.29
openclaw/openclaw#75438	imessage	channel-specific	1	-1	IMessageRpcClient.request() does not handle async stdin write errors → uncaughtException → gateway crash
openclaw/openclaw#75419	runtime	other	1	-1	[Bug]: [2026.4.29] openclaw cannot accept for any input
openclaw/openclaw#75324	discord	channel-specific	1	-1	[Discord] unresolved SecretRef at runtime despite secrets audit showing 0 unresolved
openclaw/openclaw#75420	runtime	other	1	-1	Bug: web_search can receive unresolved SecretRef for Gemini provider after 2026.4.29
openclaw/openclaw#75411	runtime	runtime	3	-3	[Bug]: Copilot 429 fail-fast fix is bypassed by session-custom stream path on 2026.4.29
openclaw/openclaw#75297	gateway	kernel	5	-5	Gateway event-loop saturation and very slow sessions.list/models.list on all tested versions after 2026.4.23;

Stats

• Base score: 60
• Evidence penalty: -24
• Vote penalty: -2
• Editorial verdict: NO

---

IsItStable.com — community-driven stability verdicts

[ri222AI]

Title:
OpenClaw 2026.4.29 embedded agent times out on openai/gpt-5.2 and gpt-5.4 while raw OpenAI Responses/SSE works

Environment:
Ubuntu 24.04 EC2
OpenClaw 2026.4.29
Gateway via systemd
Slack socket mode
OpenAI API key loaded from service env

Working:
curl /v1/models returns HTTP 200 in ~0.8s
curl /v1/responses gpt-5.2 non-streaming returns HTTP 200 in ~1.8s
curl /v1/responses stream=true emits response.output_text.delta and response.completed

Failing:
OpenClaw Slack DM triggers agent run
Run reaches stream-ready
openai/gpt-5.2 times out after ~166s
fallback openai/gpt-5.4 also times out after ~166s
Slack receives delivered error reply

Tried:
Changed primary from gpt-5.1 to gpt-5.2
Added gpt-5.4 fallback
Disabled Slack native streaming
Forced OpenAI transport=sse
Disabled openaiWsWarmup
No improvement

Expected:
OpenClaw should consume the Responses stream and deliver assistant output.

Actual:
Embedded agent reports FailoverError: LLM request timed out.