#I swapped from codex to claude as my main agent, but now i swapped back again but i get no responce

1 messages · Page 1 of 1 (latest)

open cedarBOT
#

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

austere jasperBOT
#

A “no response / just sits there” after switching back to Codex is almost always diagnosable from models status + logs. A couple common gotchas (wrong host, allowlist, or service using different config).

Can you paste the output of these (redact tokens if any show up)?

openclaw --version
openclaw gateway status
openclaw models status --probe-provider openai-codex

Then in another terminal, run:

openclaw logs --follow

…and send a Discord message to the bot once. Grab the few log lines around the Codex attempt.

#

Things that commonly cause “nothing happens” specifically when switching to Codex:

  1. OAuth was done on the wrong machine
    Codex OAuth tokens are stored on the gateway host (auth profiles). If your Gateway is on a VPS/Mac mini but you ran openclaw onboard on your laptop, the Gateway won’t have the Codex auth.

  2. Model allowlist blocks it
    If you set agents.defaults.models (an allowlist), and Codex isn’t in it, OpenClaw can reject the selection (often shows up as “Model … is not allowed” in logs / status).

  3. CLI vs service config mismatch
    openclaw gateway status will show Config (cli) vs Config (service) if they diverged—then you “changed stuff” but the running service is using something else.

Docs that match the above checks:

ionic fable
#

openclaw --version
openclaw gateway status
openclaw models status --probe-provider openai-codex
2026.2.19-2

🦞 OpenClaw 2026.2.19-2 (45d9b20) — The only bot that stays out of your training set.

│
â—‡
Service: systemd (enabled)
File logs: /tmp/openclaw/openclaw-2026-02-21.log
Command: /usr/bin/node /home/magnus/.local/share/pnpm/5/.pnpm/openclaw@2026.2.19-2_@modelcontextprotocol+sdk@1.26.0_zod@4.3.6__@napi-rs+canvas@0.1.90_74089c6f7b006658c3d14700a6051c94/node_modules/openclaw/dist/index.js gateway --port 18789
Service file: ~/.config/systemd/user/openclaw-gateway.service
Service env: OPENCLAW_GATEWAY_PORT=18789

Config (cli): ~/.openclaw/openclaw.json
Config (service): ~/.openclaw/openclaw.json

Gateway: bind=loopback (127.0.0.1), port=18789 (service args)
Probe target: ws://127.0.0.1:18789
Dashboard: http://127.0.0.1:18789/
Probe note: Loopback-only gateway; only local clients can connect.

Runtime: running (pid 296690, state active, sub running, last exit 0, reason 0)
RPC probe: ok

#

Listening: 127.0.0.1:18789
Troubles: run openclaw status
Troubleshooting: https://docs.openclaw.ai/troubleshooting

🦞 OpenClaw 2026.2.19-2 (45d9b20) — Chat automation for people who peaked at IRC.

Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : openai-codex/codex-5.3
Fallbacks (3) : anthropic/claude-sonnet-4-20250514, anthropic/claude-opus-4-6, google/gemini-2.0-flash
Image model : -
Image fallbacks (0): -
Aliases (5) : codex -> openai-codex/codex-5.3, opus -> anthropic/claude-opus-4-6, sonnet -> anthropic/claude-sonnet-4-20250514, haiku -> anthropic/claude-haiku-4-5, gemini -> google/gemini-2.0-flash
Configured models (5): openai-codex/codex-5.3, anthropic/claude-opus-4-6, anthropic/claude-sonnet-4-20250514, anthropic/claude-haiku-4-5, google/gemini-2.0-flash

#

Auth overview
Auth store : ~/.openclaw/agents/main/agent/auth-profiles.json
Shell env : off
Providers w/ OAuth/tokens (1): openai-codex (1)

  • anthropic effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | anthropic:default=sk-ant-a...tqvrCAAA | env=sk-ant-a...tqvrCAAA | source=env: ANTHROPIC_API_KEY
  • openai effective=env:sk-proj-...-39V7JoA | env=sk-proj-...-39V7JoA | source=env: OPENAI_API_KEY
  • openai-codex effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=1, token=0, api_key=0) | openai-codex:default=OAuth

Missing auth

  • google Run openclaw configure or set an API key env var.

OAuth/token status

  • openai-codex usage: 5h 100% left ⏱4h 59m · Day 100% left ⏱6d 23h
    • openai-codex:default ok expires in 10d

12:57:00 debug diagnostic {"subsystem":"diagnostic"} lane enqueue: lane=session:agent:main:main queueSize=1
12:57:00 debug diagnostic {"subsystem":"diagnostic"} lane dequeue: lane=session:agent:main:main waitMs=1 queueSize=0
12:57:00 debug diagnostic {"subsystem":"diagnostic"} lane enqueue: lane=main queueSize=1
12:57:00 debug diagnostic {"subsystem":"diagnostic"} lane dequeue: lane=main waitMs=2 queueSize=0
12:57:00 error diagnostic {"subsystem":"diagnostic"} lane task error: lane=main durationMs=1 error="Error: Unknown model: openai-codex/codex-5.3"
12:57:00 error diagnostic {"subsystem":"diagnostic"} lane task error: lane=session:agent:main:main durationMs=5 error="Error: Unknown model: openai-codex/codex-5.3"
12:57:00 error Embedded agent failed before reply: Unknown model: openai-codex/codex-5.3

austere jasperBOT
#

gateway status looks healthy (RPC probe: ok), so the next suspect is the model/provider side (Codex OAuth token missing/expired, probe hanging, or Codex requests timing out).

I don’t see the output of this one in your paste — did it hang?

openclaw models status --probe-provider openai-codex

Can you run these and paste the full output:

openclaw models status --plain
openclaw models status --probe-provider openai-codex --probe-timeout 15000 --plain
openclaw channels status --probe

Also, while reproducing (send one Discord message that “just sits there”), tail logs and paste the chunk around the attempt:

# 
openclaw logs --follow

What I’m looking for in logs is something like:

  • provider=openai-codex ... then an auth error / timeout
  • or a failover chain (“tried codex → fell back → …”) that ends in an error

If the probe output says unauthenticated / expired / refresh failed, the fix is usually just re-login on the same user account that runs the systemd user service (looks like that’s your current user already):

openclaw models auth login --provider openai-codex
openclaw gateway restart

Docs refs (for the exact flows / expectations):

#

Your models status confirms the config side is fine (default is openai-codex/codex-5.3, and it’s in the configured models list) — what’s missing is the auth/probe section that tells us whether Codex OAuth is actually usable right now.

Can you run exactly these and paste the output:

openclaw models status --probe-provider openai-codex --probe-timeout 15000 --plain
openclaw channels status --probe
openclaw status

If that Codex probe hangs or comes back with anything like 401, invalid_grant, missing credentials, etc., the most common fix is simply re-doing the Codex OAuth on the same account that runs the gateway service:

openclaw models auth login --provider openai-codex
openclaw gateway restart
#

One more common gotcha: if you logged into Codex in multiple places (OpenClaw + Codex CLI), refresh tokens can invalidate each other (“randomly logged out later”). Re-login via OpenClaw usually stabilizes it. Docs: https://docs.openclaw.ai/concepts/oauth

Once you paste those three outputs, we can tell whether it’s:

  • Codex auth broken (most likely), or
  • Discord can’t deliver messages (permissions / channel state), or
  • Codex requests timing out and falling into a bad fallback path.
#

Your Codex auth is fine (openai-codex:default ok expires in 10d), but the reason you get “no response” is in your log:

Unknown model: openai-codex/codex-5.3

So your default model string is not a valid model id in this build, and the run dies before it can reply.

Fix:

  1. List the real Codex model ids your gateway knows about:
openclaw models list

(then look for the openai-codex/... entries)

  1. Set the default to a valid one (most commonly it’s this):