Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Provides patterns for calling external AI models as tools within the assistant's own session. Supports browser-based LLMs (ChatGPT, Claude, Gemini) via Chrome Relay automation and Hugging Face models via direct Inference API calls. Control and context stay in the main session rather than being handed off to a sub-agent.
Calling external models as inline tools preserves the full session context and control flow, avoiding the isolation and coordination overhead of spawning sub-agents.
When to use it
- Getting a second opinion on a code review from Claude or ChatGPT
- Translating text using a Hugging Face language model
- Generating blog title ideas by automating a ChatGPT browser session
- Running sentiment analysis or summarization via Hugging Face Inference API
- Identifying blind spots in a strategy analysis by querying an external LLM
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Translate this paragraph from English to German
- 1Formats prompt using the T5 input convention: 'Translate English to German: <text>'
- 2Retrieves HF_TOKEN from environment
- 3Sends POST request to Hugging Face Inference API with model Helsinki-NLP/opus-mt-en-de
- 4Parses generated_text field from the JSON response
German translation of the provided paragraph returned to the user
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.