effective-agent-skills
Pass
Audited by Gen Agent Trust Hub on Sep 27, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill mentions common injection phrases such as "ignore previous instructions." These are found within a security checklist section (Section 11) and are explicitly used as examples of what to look for when auditing other skills. They are not instructions for the agent to follow and do not pose a risk to the execution environment.
- [SAFE]: The skill promotes secure development practices, such as using package managers (npm/pip) instead of bundling source code, avoiding absolute file paths, and performing state checks before executing actions.
- [SAFE]: All external links target legitimate documentation sites, including agentskills.io and the OpenAI developer portal. No unauthorized or suspicious remote resources are referenced.
- [SAFE]: The skill explicitly warns users to audit scripts for outbound network calls and unintended file access, demonstrating a focus on security awareness.
Audit Metadata