Skill Creator (Opencode)

Create new skills, modify and improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, edit, or optimize a...

This package is an edited fork of a known 'skill-creator' and is internally consistent with that purpose, but you should review and control runtime actions before use: - Review the Python scripts (especially those that call subprocess.run, lsof, os.kill, or open an HTTP server) before executing to ensure you trust them and understand what external commands they will run. - The skill expects an Opencode CLI on your machine and will invoke it (opencode run). Only allow that if you trust the Opencode binary on your system. - The eval viewer starts a small HTTP server and embeds workspace files (including base64-encoded binaries) into a page. Ensure the server binds to localhost (and is not exposed to the public internet) before opening it, and don't point it at directories containing secrets you don't want served. - The scripts may attempt to kill processes on a port (uses lsof/os.kill) — run as a user with appropriate privileges and be cautious if you run as root. - Because this is code included with the skill (no packaged install), prefer running in a sandboxed environment (non-privileged account, isolated VM/container) if you plan to execute the scripts. If you only want the guidance functionality without executing any code, you can still use the SKILL.md instructions as a human-readable workflow; executing the included scripts is optional.

Type: OpenClaw Skill Name: skill-creator-opencode Version: 1.0.1 This skill bundle is a development utility designed to create, test, and optimize OpenClaw skills using the Opencode CLI. It includes scripts for benchmarking, packaging, and a local HTML viewer (eval-viewer/generate_review.py) that runs a temporary local HTTP server for result visualization. While the skill performs subprocess execution and filesystem modifications (e.g., in scripts/run_eval.py and scripts/improve_description.py), these actions are necessary for its stated purpose of skill development and evaluation. The SKILL.md instructions explicitly direct the agent to avoid creating malware or facilitating unauthorized access, and no indicators of data exfiltration or malicious intent were identified.