#Quick start issue

1 messages · Page 1 of 1 (latest)

undone ice
#

I am using vps m7i-flex.large, trying using one-liner (curl -fsSL https://openclaw.ai/install.sh | bash) to quick start openclaw. However, when I just clicked "Quick Start", it just derectly scrolled to "Select channel (QuickStart)", and indicated that
◇ Model check ───────────────────────────────────────────────────────────────────────────╮
│ │
│ No auth configured for provider "anthropic". The agent may fail until credentials are │
│ added. │
│ │
├─────────────────────────────────────────────────────────────────────────────────────────╯
leaving me no choice to choose what model to use and enter my api key. I planned to use Open Router's api key. What should I do?

silver voidBOT
#

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

torn kilnBOT
#

That “No auth configured for provider anthropic” line is just a warning from the Quick Start path because the default model it expects is Anthropic. Quick Start doesn’t force you through model/auth setup, so it jumps straight to channel selection.

If you want to use OpenRouter, you just need to (1) add your OpenRouter key and (2) switch the default model away from anthropic/*.

0) First: what version are you on?

On the VPS run (and paste the output, no secrets):

openclaw --version
openclaw models status

Option A (recommended): set OpenRouter key in config + auto-pick a model via scan

  1. Set the key (don’t paste it into Discord):
openclaw config set env.OPENROUTER_API_KEY "sk-or-..."
# 
openclaw config set env.OPENROUTER_API_KEY "sk-or-..."
  1. See available OpenRouter models and choose one:
openclaw models list --all --provider openrouter
  1. Set it:
openclaw models set openrouter/<provider>/<model>
  1. Restart:
undone ice
#

I did the step 0:
openclaw --version
openclaw models status
while it outputted that
openclaw: command not found
openclaw: command not found

torn kilnBOT
#

That means the installer didn’t leave an openclaw binary on your current PATH (either install didn’t complete, or it installed into an npm “global bin” dir that your shell doesn’t know about yet).

Run these on the VPS and paste the output (safe; no secrets):

whoami
cat /etc/os-release | head
node -v
npm -v

npm config get prefix
ls -la "$(npm config get prefix)/bin" | head -n 50
command -v openclaw || true
echo "$PATH"

Common fixes (pick what matches what you see)

#

1) $(npm config get prefix)/bin/openclaw exists, but command -v openclaw is empty
Your PATH just needs updating. Quick test (current shell only):

export PATH="$(npm config get prefix)/bin:$PATH"
openclaw --version

Then make it permanent (bash):

echo 'export PATH="$(npm config get prefix)/bin:$PATH"' >> ~/.profile
source ~/.profile

(If you’re using zsh, use ~/.zshrc instead.)

#

2) There is no openclaw in that bin directory
Then OpenClaw likely didn’t install. On Linux the simplest install is:

npm i -g openclaw@latest

After that, re-run:

openclaw --version

3) You installed as root but are now a different user (or vice versa)
npm -g installs per-user unless you intentionally set a system prefix. The whoami + npm config get prefix output will make this obvious—then we’ll decide whether to reinstall under the user you actually want to run the gateway as.

Once openclaw --version works, we can go back to your original goal: configure OpenRouter and set the default model.

undone ice
#

For the whoami commands, this is the outputs
ubuntu
PRETTY_NAME="Ubuntu 24.04.3 LTS"
NAME="Ubuntu"
VERSION_ID="24.04"
VERSION="24.04.3 LTS (Noble Numbat)"
VERSION_CODENAME=noble
ID=ubuntu
ID_LIKE=debian
HOME_URL="https://www.ubuntu.com/"
SUPPORT_URL="https://help.ubuntu.com/"
BUG_REPORT_URL="https://bugs.launchpad.net/ubuntu/"
v22.22.0
10.9.4
/home/ubuntu/.npm-global
total 8
drwxrwxr-x 2 ubuntu ubuntu 4096 Feb 14 12:39 .
drwxrwxr-x 4 ubuntu ubuntu 4096 Feb 14 12:13 ..
lrwxrwxrwx 1 ubuntu ubuntu 41 Feb 14 12:39 openclaw -> ../lib/node_modules/openclaw/openclaw.mjs
/usr/local/sbin:/usr/local/bin:/usr/sbin:/usr/bin:/sbin:/bin:/usr/games:/usr/local/games:/snap/bin

torn kilnBOT
#

Nice — the key detail is here:

  • npm prefix: /home/ubuntu/.npm-global
  • openclaw exists at: /home/ubuntu/.npm-global/bin/openclaw
  • your PATH does not include /home/ubuntu/.npm-global/bin

So openclaw is installed, your shell just can’t find it.

Fix PATH (current shell) + verify

Run:

export PATH="$HOME/.npm-global/bin:$PATH"
openclaw --version

Make it permanent (so it works after reconnect / in new shells)

# 
echo 'export PATH="$HOME/.npm-global/bin:$PATH"' >> ~/.profile
source ~/.profile

(Then you can SSH out/in, and openclaw should still work.)

After openclaw works: set up OpenRouter

Don’t paste your key in Discord. On the VPS:

openclaw config set env.OPENROUTER_API_KEY "sk-or-..."
openclaw models scan --set-default
openclaw gateway restart

If anything still says “command not found” after the PATH change, paste:

# 
ls -la /home/ubuntu/.npm-global/bin/openclaw
command -v openclaw || true
undone ice
#

Now, let we go back to your previous suggestions. I redo your suggestion of 0) First: what version are you on?
On the VPS run (and paste the output, no secrets):
openclaw --version
openclaw models status
and here is the result:
026.2.13

🦞 OpenClaw 2026.2.13 (203b5bd) — Less middlemen, more messages.

Config : ~/.openclaw/openclaw.json
Agent dir : ~/.openclaw/agents/main/agent
Default : anthropic/claude-opus-4-6
Fallbacks (0) : -
Image model : -
Image fallbacks (0): -
Aliases (0) : -
Configured models (0): all

Auth overview
Auth store : ~/.openclaw/agents/main/agent/auth-profiles.json
Shell env : off
Providers w/ OAuth/tokens (0): -

  • openrouter effective=env:sk-or-... | env=sk-or-... | source=env: OPENROUTER_API_KEY

Missing auth

  • anthropic Run claude setup-token, then openclaw models auth setup-token or openclaw configure.

OAuth/token status

  • none
    any following suggestions? note that I only did the step 0 and didnt try further steps yet
torn kilnBOT
#

Yep — your models status output actually looks healthy for OpenRouter:

  • OpenRouter key is detected: openrouter effective=env:sk-or-...
  • The only problem is your Default model is still set to an Anthropic model (anthropic/claude-opus-4-6), so it warns that Anthropic auth is missing.

Next step is simply: switch the default model to an OpenRouter model, then restart the gateway.

Option A (easiest): let OpenClaw pick via scan

This will probe a bit (small token usage) and set a good default:

openclaw models scan --set-default
openclaw gateway restart
openclaw models status

After that, “Missing auth: anthropic” is fine, because you’re no longer using Anthropic.

#

Option B: manually pick a specific OpenRouter model

  1. List OpenRouter models:
openclaw models list --all --provider openrouter
  1. Set your default (pick one from the list output):
openclaw models set openrouter/<provider>/<model>
  1. Restart + verify:
openclaw gateway restart
openclaw models status
#

One small note about env

Your output says Shell env: off — that’s fine as long as the key is in OpenClaw config (env.OPENROUTER_API_KEY) or ~/.openclaw/.env. Just don’t rely on exporting it only in your interactive shell.

If after setting the model you still see the UI complaining about Anthropic, run and paste:

openclaw gateway status

Sometimes the service is pointed at a different config than the CLI.

undone ice
#

I just conducted Option A (recommended): set OpenRouter key in config + auto-pick a model via scan

  1. Set the key (don’t paste it into Discord):
    openclaw config set env.OPENROUTER_API_KEY "sk-or-..."
  2. Let OpenClaw pick a working OpenRouter default (interactive, and can set fallbacks):
    openclaw models scan --set-default
    and here are the results, (printed in the next message)

I want to choose GPT5.2 as my chosen LLM, what do I do?

#

Scan results: tested 31, tool ok 1, image ok 0, tool+image ok 0, image only 0
Model Tool Image Ctx Params Notes
openrouter/google/gemma-3n-e4b-it:free fail skip 8k 4b
openrouter/cognitivecomputations/dolphi... fail skip 32k 24b
openrouter/deepseek/deepseek-r1-0528:free fail skip 160k -
openrouter/google/gemma-3n-e2b-it:free fail skip 8k 2b
openrouter/nousresearch/hermes-3-llama-... fail skip 128k 405b
openrouter/openai/gpt-oss-120b:free fail skip 128k 120b
openrouter/meta-llama/llama-3.2-3b-inst... fail skip 128k 3b
openrouter/nvidia/nemotron-3-nano-30b-a... fail skip 250k 30b
openrouter/openai/gpt-oss-20b:free fail skip 128k 20b
openrouter/liquid/lfm-2.5-1.2b-instruct... fail skip 32k 1.2b
openrouter/google/gemma-3-27b-it:free fail skip 128k 27b
openrouter/liquid/lfm-2.5-1.2b-thinking... fail skip 32k 1.2b
(to be continued)

#

openrouter/z-ai/glm-4.5-air:free fail skip 128k -
openrouter/google/gemma-3-12b-it:free fail skip 32k 12b
openrouter/arcee-ai/trinity-large-previ... fail skip 128k -
openrouter/arcee-ai/trinity-mini:free fail skip 128k -
openrouter/google/gemma-3-4b-it:free fail skip 32k 4b
openrouter/nvidia/nemotron-nano-9b-v2:free fail skip 125k 9b
openrouter/openrouter/aurora-alpha fail skip 125k -
openrouter/upstage/solar-pro-3:free fail skip 125k -
openrouter/stepfun/step-3.5-flash:free fail skip 250k -
openrouter/qwen/qwen3-vl-235b-a22b-thin... fail skip 128k 235b
openrouter/qwen/qwen3-vl-30b-a3b-thinking 2.8s skip 128k 30b
openrouter/meta-llama/llama-3.3-70b-ins... fail skip 125k 70b
openrouter/mistralai/mistral-small-3.1-... fail skip 125k 24b
openrouter/qwen/qwen3-coder:free fail skip 256k 480b
openrouter/qwen/qwen3-next-80b-a3b-inst... fail skip 256k 80b
openrouter/qwen/qwen3-4b:free fail skip 40k 4b
openrouter/openrouter/free fail skip 195k -
openrouter/qwen/qwen3-235b-a22b-thinkin... fail skip 128k 235b
openrouter/nvidia/nemotron-nano-12b-v2-... fail skip 125k 12b
│
◆ Select fallback models (ordered)
│ ◻ openrouter/qwen/qwen3-vl-30b-a3b-thinking (tool 2.8s | img skip | ctx 128k | 30b)
└

torn kilnBOT
#

To make GPT‑5.2 your default while using OpenRouter, you just need to set the default model to the exact OpenRouter model id (don’t guess the spelling—copy it from the list).

1) Find the exact OpenRouter model id for GPT‑5.2

Run one of these:

openclaw models list --all --provider openrouter

Then search in the output for gpt-5.2.

If you want a one-liner filter (works if grep is available, which it is on Ubuntu):

openclaw models list --all --provider openrouter --plain | grep -i "gpt-5.2"
#

You should see something like (example format):
openrouter/openai/gpt-5.2 (but use whatever your command prints).

2) Set it as the default model

Copy/paste the exact id you found:

openclaw models set <PASTE_THE_OPENROUTER_MODEL_ID_HERE>

3) Restart + verify

openclaw gateway restart
openclaw models status

If you paste just the filtered line(s) from step 1 (the grep gpt-5.2 output), I can tell you the exact openclaw models set ... command to run.

#

openclaw models scan only scans/probes OpenRouter’s free catalog (:free), so it will not pick GPT‑5.2 (it isn’t a free model). That’s why your scan output is all ...:free and it ended up offering Qwen as a fallback.

To use GPT‑5.2 via OpenRouter, set the default model explicitly:

1) Find the exact OpenRouter id for GPT‑5.2

Run:

openclaw models list --all --provider openrouter --plain | grep -i "gpt-5.2"

2) Set it as your default model (primary)

Take the exact string from the output and run:

openclaw models set openrouter/openai/gpt-5.2

(If your models list prints a slightly different name, use that exact one instead.)

#

3) Restart + verify

openclaw gateway restart
openclaw models status

Important note: don’t use the shorthand gpt here unless you’ve also configured the OpenAI provider directly — gpt typically resolves to openai/gpt-5.2, not openrouter/....

If you paste the output of the grep -i "gpt-5.2" line, I’ll tell you the exact openclaw models set ... command to run for your catalog.

undone ice
#

Finally, I'd like to choose haiku 3.5 as my LLM for its financial merrit. After my openclaw models set <PASTE_THE_OPENROUTER_MODEL_ID_HERE> command, here's the results. Did it work well?
🦞 OpenClaw 2026.2.13 (203b5bd) — IPC, but it's your phone.

Config overwrite: /home/ubuntu/.openclaw/openclaw.json (sha256 4cb78452263e6046709782137385540ca0ff3ea00a16247c11465dc843e5ab03 -> 95d04c472e4608e59dea6b637d3e9c85d8f86804499424abe9e48481527b31ce, backup=/home/ubuntu/.openclaw/openclaw.json.bak)
Updated ~/.openclaw/openclaw.json
Default model: openrouter/anthropic/claude-3.5-haiku
ubuntu@ip-172-31-36-86:~$

torn kilnBOT
#

Yep — that output is exactly what you want to see.

  • Updated ~/.openclaw/openclaw.json
  • Default model: openrouter/anthropic/claude-3.5-haiku

So the config change worked. The only remaining step is making sure the Gateway process picks up the new config.

Do this next

openclaw gateway restart
openclaw models status

In models status you should now see:

  • Default = openrouter/anthropic/claude-3.5-haiku
  • OpenRouter auth present (from env.OPENROUTER_API_KEY)