Create Agent Skills
v1.0.0Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.
⭐ 3· 3.3k·24 current·24 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
high confidencePurpose & Capability
The name/description (skill authoring guide) matches the included files: SKILL.md, reference docs, and scripts to initialize, validate, and package skills. Nothing in the repo asks for cloud credentials, unrelated binaries, or system-wide config access that would be disproportionate.
Instruction Scope
SKILL.md is a purely authoring/documentation guide and does not instruct the agent to read arbitrary system files, exfiltrate data, or call unexpected external endpoints. It does recommend including scripts and references in skills; those scripts are present and self-contained (create files, validate content, zip a folder). As a caution: the scripts are intended to be executed by a user/environment and could modify the filesystem (create skill folders, write files, create .skill archives) — review before running.
Install Mechanism
There is no install specification; this is an instruction- and script-bundled skill. No downloads or external installers are performed by the repo itself. The README notes a dependency on PyYAML/Python 3.11+, which is expected for the included Python validator.
Credentials
The skill declares no required environment variables, credentials, or config paths. The included scripts operate on local paths and SKILL.md contents only. No secrets are requested by the skill itself; however, if you package a skill folder, any files in that folder (including secrets) will be included—see guidance below.
Persistence & Privilege
The skill does not request permanent/always-on presence and does not modify other skills' configuration. It is user-invocable and uses the normal autonomous invocation defaults; nothing indicates elevated or unusual persistence privileges.
Assessment
This skill is coherent and matches its stated purpose, but take these practical precautions before using or running the bundled scripts: 1) Review the Python scripts (init_skill.py, quick_validate.py, package_skill.py) locally to ensure they meet your expectations — they perform filesystem operations (create dirs/files, read SKILL.md, create zip archives). 2) Install required runtime deps (Python 3.11+, PyYAML) in an isolated environment if you plan to run the validator/packager. 3) Do not include secrets or private keys in a skill folder you plan to package — the packager will include all files under the skill directory. 4) Complete TODOs in SKILL.md before packaging (the validator will reject TODO placeholders). 5) If you need stronger assurance, run the scripts in a sandbox or review the archive contents before distributing. Overall, the package appears benign and appropriate for its stated role.Like a lobster shell, security has layers — review code before you run it.
latestvk975cpxq5fheppbvfra4khfjbh80erfm
License
MIT-0
Free to use, modify, and redistribute. No attribution required.