Skill Creator

PassAudited by ClawScan on May 1, 2026.

Overview

This is a coherent skill-authoring helper, with normal cautions that its helpers can write persistent skill files and package local directory contents when invoked.

This skill looks reasonable for creating and maintaining AgentSkills. Before using it, make sure helper scripts run only in the intended skill directory, inspect any generated .skill package before sharing, and review new or edited SKILL.md files because they can influence future agent behavior.

Findings (3)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

What this means

Using the helper with the wrong target path could create or modify files in an unintended workspace.

Why it was flagged

The initializer is designed to create directories and files under a caller-provided path. That is expected for this skill, but it is local filesystem mutation.

Skill content
init_skill.py <skill-name> --path <path> [--resources scripts,references,assets] [--examples]; resource_dir.mkdir(exist_ok=True)
Recommendation

Run the helper only against the intended skill workspace and review generated files before installing or sharing the skill.

What this means

If a skill folder accidentally contains private files, they could be bundled into the .skill archive and later shared.

Why it was flagged

The packager recursively includes regular files from the skill directory into the output archive, while excluding only selected directories and symlinks.

Skill content
for file_path in skill_path.rglob("*"): ... zipf.write(file_path, arcname)
Recommendation

Inspect the skill directory and the generated archive before distribution; keep secrets and unrelated files out of skill folders.

What this means

Overly broad or incorrect skill metadata could cause future agents to load the wrong instructions for unrelated tasks.

Why it was flagged

The skill teaches creation and editing of SKILL.md metadata that controls future skill triggering and context loading.

Skill content
These are the only fields that Codex reads to determine when the skill gets used
Recommendation

Review generated SKILL.md descriptions and referenced materials carefully before installing them as persistent skills.