writing-skills

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill explicitly instructs the use of 'persuasion principles' such as Authority and Commitment to influence agent behavior. It advocates for imperative language (e.g., 'YOU MUST', 'No exceptions') to override an agent's tendency to rationalize skipping steps. While these techniques are similar to prompt reinforcement used in jailbreaking research, they are applied here transparently for the intended purpose of enforcing technical discipline and workflow compliance.
  • [COMMAND_EXECUTION]: The documentation references the use of local system commands including bash, python, wc, and grep, as well as version control operations with git. These are presented as standard tools for managing skill files, measuring token efficiency, and executing local utility scripts.
  • [EXTERNAL_DOWNLOADS]: The skill mentions common, well-known software libraries such as pdfplumber, pypdf, and docx-js from established package registries (PyPI, NPM). These are documented as dependencies for specific processing tasks (e.g., PDF extraction) rather than as suspicious remote code sources.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 05:18 PM