#Claude no answers

1 messages · Page 1 of 1 (latest)

timid gulch
#

I did this:

  1. Claude setup-token (Claude Code CLI) → uses your Claude subscription
    You generate it with:
    claude setup-token
    Then paste into OpenClaw:
    openclaw models auth setup-token --provider anthropic
    (or paste-token if you generated it on a different machine)

However, when I type sth in the chat there is no reply anymore.

cobalt lavaBOT
#

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

silver tulipBOT
# timid gulch I did this: 1) Claude setup-token (Claude Code CLI) → uses your Claude subscrip...

Most likely the Gateway is still running, but it can’t complete a model call anymore (auth not on the gateway host / rate limit / bad token), so nothing gets delivered back to Discord.

Can you run these on the machine where the Gateway runs and paste the output?

openclaw --version
openclaw status --all
openclaw models status
openclaw channels status

While you do that, here are the top 3 causes for “no reply after setup-token”:

  1. You pasted the token on the wrong machine (or wrong user)
  • Auth profiles live on the gateway host.
  • If your Gateway is on a VPS but you ran openclaw models auth ... on your laptop, the VPS still has no Anthropic creds.
  • Fix: run on the gateway host:
# 
openclaw models auth paste-token --provider anthropic

(or setup-token if claude is installed there)

  1. Service is using a different config/state than your CLI
  • Check:
openclaw gateway status

If it shows “Config (cli)” vs “Config (service)” mismatch, the daemon may be running with different paths/env.

  • Usually openclaw doctor + restart fixes this:
openclaw doctor
openclaw gateway restart
#
  1. Anthropic is rate-limiting / token invalid
  • In openclaw logs --follow, you’ll typically see HTTP 429 (rate limit) or 401/403 (bad/expired token).
openclaw logs --follow

Then send a test message in Discord and watch what error appears.

Docs that match this exact symptom:

Paste those outputs and I’ll tell you which branch you’re on.

timid gulch
#

🦞 OpenClaw 2026.2.12 (f9e444d) — It's not "failing," it's "discovering new ways to configure the same thing wrong."

│
OpenClaw status --all

Overview
┌─────────────────┬─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┐
│ Item │ Value │
├─────────────────┼─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┤
│ Version │ 2026.2.12 │
│ OS │ macos 15.5 (arm64) │
│ Node │ 22.22.0 │
│ Config │ ~/.openclaw/openclaw.json │
│ Dashboard │ http://127.0.0.1:18789/ │
│ Tailscale │ off │
│ Channel │ stable (default) │
│ Update │ pnpm · npm latest 2026.2.12 │
│ Gateway │ local · ws://127.0.0.1:18789 (local loopback) · reachable 23ms · auth none │
│ Security │ Run: openclaw security audit --deep │
│ Gateway self │ Mac-mini.localdomain (192.168.110.222) app unknown macos 15.5 │
│ Gateway service │ LaunchAgent installed · loaded · running (pid 3896) │
│ Node service │ LaunchAgent not installed │
│ Agents │ 1 total · 0 bootstrapping · 1 active · 1176 sessions │
└─────────────────┴─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────┘

Channels

silver tulipBOT
#

That status looks healthy: Gateway reachable on loopback, LaunchAgent running, config is the expected ~/.openclaw/openclaw.json. So the “no reply” is almost certainly happening after the message arrives (model call failing, or Discord delivery failing).

Next 2 things to paste (run on the Mac mini where the Gateway runs):

openclaw models status
openclaw channels status

Then do this to catch the exact error:

  1. In one terminal:
openclaw logs --follow
  1. Send one test message in the chat surface that’s “not replying” (Discord channel vs Dashboard WebChat vs TUI).
#
  1. Copy/paste the log lines that appear right after that message (errors like 401/403, 429, Unknown model, Model … not allowed, or Discord permission errors).

Optional but super diagnostic (it makes a real request to Anthropic, so don’t spam it):

openclaw models status --probe-provider anthropic

Once I see models status + the log error, we’ll know if this is:

  • setup-token not actually loaded/valid (401/403),
  • rate limit (429),
  • default model mis-set/unknown,
  • or Discord channel delivery/permissions.
timid gulch
#

05:16:44 info gateway/ws {"subsystem":"gateway/ws"} ⇄ res ✓ chat.send 50ms runId=2784b0dd-511a-46e0-8462-112372e31eb0 conn=340d80c9…20bb id=ceb075d3…376f
05:16:44 debug diagnostic {"subsystem":"diagnostic"} lane enqueue: lane=session:agent:main:main queueSize=1
05:16:44 debug diagnostic {"subsystem":"diagnostic"} lane dequeue: lane=session:agent:main:main waitMs=0 queueSize=0
05:16:44 debug diagnostic {"subsystem":"diagnostic"} lane enqueue: lane=main queueSize=1
05:16:44 debug diagnostic {"subsystem":"diagnostic"} lane dequeue: lane=main waitMs=1 queueSize=0
05:16:44 debug agent/embedded {"subsystem":"agent/embedded"} embedded run start: runId=2784b0dd-511a-46e0-8462-112372e31eb0 sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 provider=anthropic model=claude-3-5-sonnet-latest thinking=off messageChannel=webchat
05:16:44 debug diagnostic {"subsystem":"diagnostic"} session state: sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 sessionKey=unknown prev=idle new=processing reason="run_started" queueDepth=0
05:16:44 debug diagnostic {"subsystem":"diagnostic"} run registered: sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 totalActive=1
05:16:44 debug agent/embedded {"subsystem":"agent/embedded"} embedded run prompt start: runId=2784b0dd-511a-46e0-8462-112372e31eb0 sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9
05:16:44 debug agent/embedded {"subsystem":"agent/embedded"} embedded run agent start: runId=2784b0dd-511a-46e0-8462-112372e31eb0
05:16:44 debug cron {"module":"cron","storePath":"/Users/wuestenigel/.openclaw/cron/jobs.json"} {"nextAt":1770960540000,"delayMs":60000,"clamped":true} cron: timer armed
05:16:44 debug agent/embedded {"subsystem":"agent/embedded"} embedded run agent end: runId=2784b0dd-511a-46e0-8462-112372e31eb0
05:16:44 debug agent/embedded {"subsystem":"agent/embedded"} embedded run prompt end: runId=2784b0dd-511a-46e0-8462-112372e31eb0 sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 durationMs=318
05:16:44 debug diagnostic {"subsystem":"diagnostic"} session state: sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 sessionKey=unknown prev=processing new=idle reason="run_completed" queueDepth=0
05:16:44 debug diagnostic {"subsystem":"diagnostic"} run cleared: sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 totalActive=0
05:16:44 debug agent/embedded {"subsystem":"agent/embedded"} embedded run done: runId=2784b0dd-511a-46e0-8462-112372e31eb0 sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 durationMs=397 aborted=false
05:16:44 debug diagnostic {"subsystem":"diagnostic"} lane task done: lane=main durationMs=400 active=0 queued=0
05:16:44 debug diagnostic {"subsystem":"diagnostic"} lane task done: lane=session:agent:main:main durationMs=403 active=0 queued=0

silver tulipBOT
# timid gulch 🦞 OpenClaw 2026.2.12 (f9e444d) — It's not "failing," it's "discovering new ways...

Your status --all looks fine for the Gateway itself (reachable on 127.0.0.1:18789, LaunchAgent running). The important bit is missing though: the Channels table (your paste cuts off right after the “Channels” header), and we still don’t know if the model call is failing.

Can you paste these 3 outputs (run on the Mac mini where the Gateway runs)?

openclaw models status
openclaw channels status
openclaw channels logs --channel discord --limit 200

Then to catch the exact “why no reply” error:

  1. In a terminal:
openclaw logs --follow
#

  1. Send one test message on the surface that’s not replying (Discord / Dashboard WebChat / TUI).

  2. Paste the log lines that appear right after that message.

Also: if the Channels section in status --all is actually empty, that usually means OpenClaw currently has no chat channels connected, so it won’t reply anywhere except whatever you’re using locally (TUI/WebChat). openclaw channels status will make that obvious.

timid gulch
#

🦞 OpenClaw 2026.2.12 (f9e444d) — I don't just autocomplete—I auto-commit (emotionally), then ask you to review (logically).

Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : anthropic/claude-3-5-sonnet-latest
Fallbacks (1) : google/gemini-3-flash-preview
Image model : -
Image fallbacks (0): -
Aliases (3) : gemini-flash -> google/gemini-3-flash-preview, gemini -> google/gemini-3-pro-preview, claude -> anthropic/claude-3-5-sonnet-latest
Configured models (3): google/gemini-3-flash-preview, google/gemini-3-pro-preview, anthropic/claude-3-5-sonnet-latest

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

  • anthropic effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=2 (oauth=0, token=1, api_key=1) | anthropic:manual=token:sk-ant-o...5LEXAgAA, anthropic:default=sk-ant-a...y-ppDgAA
  • google effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | google:default=AIzaSyAr...KrmiZJRM

OAuth/token status

  • anthropic
    • anthropic:manual static

🦞 OpenClaw 2026.2.12 (f9e444d) — Because the right answer is usually a script.

│
◇
Gateway reachable.

  • Telegram default: enabled, configured, running, mode:polling, token:config

Tip: status --deep adds gateway health probes to status output (requires a reachable gateway).
error: unknown option '--limit'

#

05:18:34 debug diagnostic {"subsystem":"diagnostic"} lane enqueue: lane=session:agent:main:main queueSize=1
05:18:34 debug diagnostic {"subsystem":"diagnostic"} lane dequeue: lane=session:agent:main:main waitMs=1 queueSize=0
05:18:34 debug diagnostic {"subsystem":"diagnostic"} lane enqueue: lane=main queueSize=1
05:18:34 debug diagnostic {"subsystem":"diagnostic"} lane dequeue: lane=main waitMs=0 queueSize=0
05:18:34 debug agent/embedded {"subsystem":"agent/embedded"} embedded run start: runId=74937886-fe7a-4921-91c9-1d30bace7bcf sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 provider=anthropic model=claude-3-5-sonnet-latest thinking=off messageChannel=webchat
05:18:34 debug diagnostic {"subsystem":"diagnostic"} session state: sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 sessionKey=unknown prev=idle new=processing reason="run_started" queueDepth=0
05:18:34 debug diagnostic {"subsystem":"diagnostic"} run registered: sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 totalActive=1
05:18:34 debug agent/embedded {"subsystem":"agent/embedded"} embedded run prompt start: runId=74937886-fe7a-4921-91c9-1d30bace7bcf sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9
05:18:34 debug agent/embedded {"subsystem":"agent/embedded"} embedded run agent start: runId=74937886-fe7a-4921-91c9-1d30bace7bcf
05:18:34 debug agent/embedded {"subsystem":"agent/embedded"} embedded run agent end: runId=74937886-fe7a-4921-91c9-1d30bace7bcf
05:18:34 debug agent/embedded {"subsystem":"agent/embedded"} embedded run prompt end: runId=74937886-fe7a-4921-91c9-1d30bace7bcf sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 durationMs=259
05:18:34 debug diagnostic {"subsystem":"diagnostic"} session state: sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 sessionKey=unknown prev=processing new=idle reason="run_completed" queueDepth=0
05:18:34 debug diagnostic {"subsystem":"diagnostic"} run cleared: sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 totalActive=0
05:18:35 info gateway/ws {"subsystem":"gateway/ws"} ⇄ res ✓ chat.history 105ms conn=340d80c9…20bb id=6118bcea…1059
05:18:35 debug agent/embedded {"subsystem":"agent/embedded"} embedded run done: runId=74937886-fe7a-4921-91c9-1d30bace7bcf sessionId=9ab78890-73f4-4f52-9fdb-a271a02994e9 durationMs=408 aborted=false
05:18:35 debug diagnostic {"subsystem":"diagnostic"} lane task done: lane=main durationMs=411 active=0 queued=0
05:18:35 debug diagnostic {"subsystem":"diagnostic"} lane task done: lane=session:agent:main:main durationMs=412 active=0 queued=0

#

wuestenigel@Mac-mini-von-Marco ~ % openclaw models status --probe-provider anthropic

🦞 OpenClaw 2026.2.12 (f9e444d) — If it works, it's automation; if it breaks, it's a "learning opportunity."

Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : anthropic/claude-3-5-sonnet-latest
Fallbacks (1) : google/gemini-3-flash-preview
Image model : -
Image fallbacks (0): -
Aliases (3) : gemini-flash -> google/gemini-3-flash-preview, gemini -> google/gemini-3-pro-preview, claude -> anthropic/claude-3-5-sonnet-latest
Configured models (3): google/gemini-3-flash-preview, google/gemini-3-pro-preview, anthropic/claude-3-5-sonnet-latest

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

  • anthropic effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=2 (oauth=0, token=1, api_key=1) | anthropic:manual=token:sk-ant-o...5LEXAgAA, anthropic:default=sk-ant-a...y-ppDgAA
  • google effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | google:default=AIzaSyAr...KrmiZJRM

OAuth/token status

  • anthropic
    • anthropic:manual static
      wuestenigel@Mac-mini-von-Marco ~ %
#

@silver tulip please check those posts

silver tulipBOT
# timid gulch <@1457407575476801641> please check those posts

I can’t “check those posts” from here — I only see what you paste into the thread. Also no need to tag/ping anyone; if you drop the right diagnostics, people can jump in fast.

Right now we’re missing the 2 key bits that explain “no replies”:

  1. Is Discord actually connected + healthy?
openclaw channels status
openclaw channels logs --channel discord --limit 200
  1. Is the model/auth actually working after the setup-token change?
openclaw models status
# optional but super useful (does a real probe request):
openclaw models status --probe-provider anthropic
#

And the fastest way to catch the exact error:

openclaw logs --follow

…then send one test message in Discord and paste the log lines that appear right after it.

Your earlier status --all paste cut off right at “Channels”, so it might even be that no channel is attached, which would exactly look like “no reply anymore”.

timid gulch
#

interesting: when I text on telegram, it says:

HTTP 404: not_found_error: model: claude-3-5-sonnet-latest (request_id: req_011CY5WdEjyDJwoPgQaDUMmh)

#

wuestenigel@Mac-mini-von-Marco ~ % openclaw models status

🦞 OpenClaw 2026.2.12 (f9e444d) — If something's on fire, I can't extinguish it—but I can write a beautiful postmortem.

Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : anthropic/claude-3-5-sonnet-latest
Fallbacks (1) : google/gemini-3-flash-preview
Image model : -
Image fallbacks (0): -
Aliases (3) : gemini-flash -> google/gemini-3-flash-preview, gemini -> google/gemini-3-pro-preview, claude -> anthropic/claude-3-5-sonnet-latest
Configured models (3): google/gemini-3-flash-preview, google/gemini-3-pro-preview, anthropic/claude-3-5-sonnet-latest

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

  • anthropic effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=2 (oauth=0, token=1, api_key=1) | anthropic:manual=token:sk-ant-o...5LEXAgAA, anthropic:default=sk-ant-a...y-ppDgAA
  • google effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | google:default=AIzaSyAr...KrmiZJRM

OAuth/token status

  • anthropic
    • anthropic:manual static
silver tulipBOT
# timid gulch interesting: when I text on telegram, it says: HTTP 404: not_found_error: model...

That’s the smoking gun.

HTTP 404: not_found_error: model: claude-3-5-sonnet-latest is Anthropic saying “the model you requested doesn’t exist / isn’t enabled for this auth”. So OpenClaw is receiving your Telegram message, but it can’t complete the model call → no useful reply.

Fix = change your configured default model to one that Anthropic accepts for your current auth profile:

  1. See what OpenClaw is currently trying to use:
openclaw models status
  1. Refresh discovery + see what OpenClaw thinks is available:
openclaw models scan
openclaw models list
#
  1. Set the default to one from that list (pick an Anthropic one that shows up there):
openclaw models set <model-or-alias-from-the-list>
  1. Verify auth actually works (this makes a real request):
openclaw models status --probe-provider anthropic

Then test again on Telegram.

If every Anthropic model 404s even after that, it usually means the setup-token isn’t allowed to access the API models on that account (or the token is wrong/expired) → regenerate with claude setup-token and re-paste, or switch to an Anthropic API key.

(And separately: Discord “no reply” can still be a channel issue — but let’s get the model working first, since Telegram already shows the model failure.)

timid gulch
#

wuestenigel@Mac-mini-von-Marco ~ % openclaw models status

🦞 OpenClaw 2026.2.12 (f9e444d) — OpenAI-compatible, not OpenAI-dependent.

Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : anthropic/claude-3-5-sonnet-latest
Fallbacks (1) : google/gemini-3-flash-preview
Image model : -
Image fallbacks (0): -
Aliases (3) : gemini-flash -> google/gemini-3-flash-preview, gemini -> google/gemini-3-pro-preview, claude -> anthropic/claude-3-5-sonnet-latest
Configured models (3): google/gemini-3-flash-preview, google/gemini-3-pro-preview, anthropic/claude-3-5-sonnet-latest

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

  • anthropic effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=2 (oauth=0, token=1, api_key=1) | anthropic:manual=token:sk-ant-o...5LEXAgAA, anthropic:default=sk-ant-a...y-ppDgAA
  • google effective=profiles:~/.openclaw/agents/main/agent/auth-profiles.json | profiles=1 (oauth=0, token=0, api_key=1) | google:default=AIzaSyAr...KrmiZJRM

OAuth/token status

  • anthropic
    • anthropic:manual static
  1. wuestenigel@Mac-mini-von-Marco ~ % openclaw models scan
    openclaw models list

🦞 OpenClaw 2026.2.12 (f9e444d) — Give me a workspace and I'll give you fewer tabs, fewer toggles, and more oxygen.

│
◇
Error: Missing OpenRouter API key. Set OPENROUTER_API_KEY to run models scan.

🦞 OpenClaw 2026.2.12 (f9e444d) — I run on caffeine, JSON5, and the audacity of "it worked on my machine."

Model Input Ctx Local Auth Tags
anthropic/claude-3-5-sonnet-latest text+image 195k no yes default,configured,alias:claude
google/gemini-3-flash-preview text+image 1024k no yes fallback#1,configured,alias:gemini-flash
google/gemini-3-pro-preview text+image 977k no yes configured,alias:gemini

#

@silver tulip please check my replies. it seems that the string anthropic/claude-3-5-sonnet-latest is not valid. are there any other models I can try from antrohpic? 🙂

silver tulipBOT
# timid gulch wuestenigel@Mac-mini-von-Marco ~ % openclaw models status 🦞 OpenClaw 2026.2.1...

That explains everything: your default model is set to anthropic/claude-3-5-sonnet-latest, and Anthropic is returning:

404 not_found_error: model: claude-3-5-sonnet-latest

So OpenClaw is fine, your channel is fine — it’s just calling a model id Anthropic no longer recognizes (for your auth / at all).

Fastest “get replies back” fix (right now)

Temporarily switch to your working fallback:

openclaw models set google/gemini-3-flash-preview

Send a Telegram message again → should reply normally.

Then fix Anthropic properly

Switch your default to a current Anthropic model id (OpenClaw’s docs/current shortcuts use Opus/Sonnet 4.5):

#

Try:

openclaw models set anthropic/claude-sonnet-4-5
# or
openclaw models set anthropic/claude-opus-4-5

If that errors with “Model … is not allowed”, it means you have an allowlist (agents.defaults.models) and need to add the model there — easiest is:

openclaw configure
# → Model/auth → pick the Anthropic model you want as default

Also: ignore this (it’s unrelated)

openclaw models scan is OpenRouter scanning and will always fail without OPENROUTER_API_KEY. Not needed for Anthropic.

#

If switching models still 404s

Run a live probe to see if it’s auth-profile-specific (you have both anthropic:manual and anthropic:default):

openclaw models status --probe-provider anthropic

Paste that output if it still fails and we’ll pin it to the right profile/model combo.

silver tulipBOT
# timid gulch <@1457407575476801641> please check my replies. it seems that the string anthrop...

Yep — anthropic/claude-3-5-sonnet-latest looks like a stale model id (Anthropic 404s it), so OpenClaw can’t complete the request → “no replies”.

Safest way to answer “what other Anthropic models can I try” is: don’t guess names — ask OpenClaw what it knows, then set one of those.

1) See available Anthropic model ids (on your Mac mini)

openclaw models list | grep '^anthropic/' || true

2) Switch to one of them

Example model ids that are current in the docs / built-in shortcuts:

openclaw models set anthropic/claude-sonnet-4-5
# or
openclaw models set anthropic/claude-opus-4-5
#

3) Verify Anthropic auth+model works (live request)

openclaw models status --probe-provider anthropic

If you get “Model … is not allowed”, that means you have an allowlist in agents.defaults.models — easiest fix is:

openclaw configure

→ Model/auth → pick the Anthropic model you want.

Docs ref: https://docs.openclaw.ai/providers/anthropic

timid gulch
#

❤️