#How to install dependencies on built-in skills?

12 messages · Page 1 of 1 (latest)

wooden flare
#

Hi Krill, when choosing a package manager on onboarding I chose pnpm but it wasn't installed yet so dependencies weren't installed. Well, how do I install them now?

maiden kestrelBOT
#

Yep, you can fix that after onboarding.

If you mean built-in skill dependencies that got skipped:

openclaw skills check

Then either:

  • keep pnpm: install it first, for example
    npm install -g pnpm
  • or switch skill installs back to npm
# 
  openclaw config set skills.install.nodeManager npm

After that, run:

openclaw onboard

and go through the Skills step again. On macOS, the Skills settings screen uses the same install flow.

One important gotcha: openclaw skills install ... is for ClawHub skills, not bundled skill dependencies.

If you actually meant repo/source dependencies, then once pnpm exists:

wooden flare
#

It asked to hatch my bot however I don't understand what's this and I only want to install the dependencies. Isn't there any way to install them manually?

maiden kestrelBOT
#

Yep, that “hatch your bot” thing is just the first-run agent/bootstrap flow, not the dependency installer. If you only want deps, choose “Do this later.”

A couple important bits:

  • There is not really a dedicated CLI command just for bundled skill dependency installs right now.
  • The supported install paths for those are basically:
    • onboarding
    • the Skills settings UI (macOS app)

If you want to avoid pnpm entirely, switch the installer back to npm first:

openclaw config set skills.install.nodeManager npm

Then rerun onboarding, pick Do this later for hatch, and use the Skills step.

#

If you want to do it manually, the best you can do is:

openclaw skills check
openclaw skills list --verbose

That shows which skills are missing requirements, then you install those binaries/packages yourself. It varies per skill, because skill installers can use brew / node / go / uv / download, not just pnpm.

So short version:

  • “Hatch” = ignore / do later
  • No single manual openclaw ... command for bundled skill deps
  • Best workaround: switch to npm + rerun the Skills step

Docs: https://docs.openclaw.ai/help/faq https://docs.openclaw.ai/tools/skills https://docs.openclaw.ai/tools/skills-config

If you want, paste the output of openclaw skills list --verbose and I can tell you exactly which deps you need.

wooden flare
#

Is there a way to automatically load all available models from LM Studio or I have to manually define them? If I have to manually define them, can you give me the config for defining them and making them visible in the UI. I'm running OpenClaw 2026.4.14 (323493f).

maiden kestrelBOT
#

I don’t see an LM Studio auto-discovery path in the current docs/build. LM Studio is treated as a custom OpenAI-compatible provider, so for now you should define the models manually.

The key part for UI visibility is:

  • models.providers.lmstudio.models[] = define the LM Studio catalog
  • agents.defaults.models = allowlist + aliases for /model and pickers

So if you want them selectable in the UI, add them to both.

Example:

{
  agents: {
    defaults: {
      model: {
# 
        primary: "lmstudio/<model-id-1>"
      },
      models: {
        "lmstudio/<model-id-1>": { alias: "Local 1" },
        "lmstudio/<model-id-2>": { alias: "Local 2" }
      }
    }
  },

  models: {
    mode: "merge",
    providers: {
      lmstudio: {
        baseUrl: "http://127.0.0.1:1234/v1",
        apiKey: "lmstudio",
# 
        api: "openai-responses",
        models: [
          {
            id: "<model-id-1>",
            name: "Local Model 1",
            reasoning: false,
            input: ["text"],
            cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
            contextWindow: 131072,
            maxTokens: 8192
          },
          {
            id: "<model-id-2>",
            name: "Local Model 2",
            reasoning: false,
# 
            input: ["text"],
            cost: { input: 0, output: 0, cacheRead: 0, cacheWrite: 0 },
            contextWindow: 131072,
            maxTokens: 8192
          }
        ]
      }
    }
  }
}

A few notes:

  • Replace \<model-id-1> / \<model-id-2> with the exact IDs from LM Studio’s:
    http://127.0.0.1:1234/v1/models