Skill Creator
v1.0.0Create 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...
⭐ 0· 71·0 current·0 all-time
MIT-0
Download zip
LicenseMIT-0 · Free to use, modify, and redistribute. No attribution required.
Security Scan
OpenClaw
Benign
medium confidencePurpose & Capability
The name/description match the content: this is a meta-skill for creating and evaluating skills. It does reference running evaluation tooling and an agent named "claude-with-access-to-the-skill" which is coherent for an authoring workflow, but those tools are external and not provided with the skill.
Instruction Scope
SKILL.md stays on-topic (interviewing users, drafting SKILL.md, writing tests, running evals). However it explicitly instructs the agent to run external scripts (e.g., eval-viewer/generate_review.py) and to invoke "claude-with-access-to-the-skill" and possible subagents — those are not included in the bundle and would require executing external code or services. It does not request reading arbitrary system files or secrets in its instructions, but it gives broad discretion to run evals and research via subagents.
Install Mechanism
No install spec and no code files are included (instruction-only). That minimizes on-disk installation risk, but also means referenced scripts/tools are external dependencies you must supply or trust.
Credentials
The skill declares no required environment variables, credentials, or config paths. Nothing in SKILL.md asks for secrets. The only environment implication is that it expects external tooling and runtime access to evaluation scripts and subagents.
Persistence & Privilege
always:false and default invocation settings are used. The skill does not request persistent installation or system-wide changes in its instructions.
Assessment
This skill is an instruction-only authoring tool and appears internally consistent, but it references specific external scripts (e.g., eval-viewer/generate_review.py) and a runtime named "claude-with-access-to-the-skill" that are not bundled. Before using it, verify you have or trust those tools and inspect any referenced scripts before running them. If you plan to allow autonomous invocation, be cautious: the agent may run external evals or subagents on your behalf — ensure they don't have access to secrets or systems you wouldn't want accessed. If you need to use this skill, consider (1) supplying vetted implementations of referenced scripts, (2) running first in a sandbox, and (3) limiting what subagents/tools the agent may call.Like a lobster shell, security has layers — review code before you run it.
latestvk970s665nayhnwcxrm839g82k983gkhy
License
MIT-0
Free to use, modify, and redistribute. No attribution required.