About this skill

Problem: skills can remain prompt-level

skill-distill addresses cases where an existing skill describes a workflow but does not yet become a standalone executable artifact. It converts the target skill into a shell script, preserving the relevant model configuration so engineers do not have to reassemble prompts, parameters, and invocation details by hand.

How it works and where it fits

The key steps are reading the target skill, capturing the current model configuration, generating a self-contained script, and saving it under ~/.openclaw/workspace/tools/. The script should remain executable, embed the model configuration as a JSON string in a header comment, and minimize external dependencies. When an LLM call is required, the skill handles API key setup: environment variables are preferred, a ~/.api_keys.sh file is also supported, and a check_api_key() function should run before calling the model.

It is useful for turning an existing skill into a lightweight tool, debugging helper, or internal automation entry point. It is not a substitute for a full service framework, access governance, or production deployment pipeline.

Screenshots

Use Cases

Turn an existing skill's invocation steps into a standalone shell tool
Embed model configuration in the script and generate an API key check function
Write the generated script into ~/.openclaw/workspace/tools/ and make it executable
Debug an LLM API-dependent skill by validating credentials before invoking the model

Best For

Engineers maintaining an internal skill library who want prompt workflows as CLI tools
Backend engineers debugging automation who need API key checks before LLM calls
DevOps engineers delivering skills to local environments with minimal dependencies
AI application engineers wrapping model calls and preserving JSON config in script headers