increase timeout while waiting for slow prompt processing of local llm

[sinful marten]

processing prompt takes long on my local machine and openclaw seems to timeout after like 10 minutes

this it the log from the gateway
Feb 04 18:52:15 logbot node[1067]: 2026-02-04T17:52:15.841Z [agent/embedded] embedded run timeout: runId=94c86bd0-c80b-45fe-9d1c-ead17e0febec sessionId=df4e5aa4-a770-4416-873c-c3f923ae653f timeoutMs=600000

lm studio log shows that the client disconnected

can we add/is there a setting for openclaw.json to increase the timeout?

[faint idol]

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

[inland lava]

Yep — that log is the embedded agent run timeout, and the default is 600s = 10 minutes.

Add / bump this in ~/.openclaw/openclaw.json (it’s JSON5):

{
  agents: {
    defaults: {
      // default is 600 (10 min)
      timeoutSeconds: 1800, // 30 min (pick what you need)
    },
  },
}

Then restart the Gateway (or the service) so it picks it up.

Docs: https://docs.openclaw.ai/gateway/configuration

Two quick notes:

• If you set it too high, a truly-stuck local model call will tie up a run for a long time.
• LM Studio saying “client disconnected” is consistent with OpenClaw aborting at the timeout, so increasing this should stop that specific disconnect.

Also: what version are you on? (openclaw --version)