writing-skills

Pass

Audited by Gen Agent Trust Hub on Oct 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill includes documentation on 'Persuasion Principles' (Authority, Scarcity, Commitment) designed to enforce compliance and prevent agents from rationalizing away rules. While these are presented as tools for quality assurance, they represent a methodology for overriding agent decision-making logic.
  • [COMMAND_EXECUTION]: The 'render-graphs.js' utility script uses 'child_process.execSync' to run the 'dot' command from Graphviz. This is a functional requirement for rendering diagrams but grants the skill shell execution capabilities.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface involving the processing of external data. 1. Ingestion points: 'render-graphs.js' reads and parses 'SKILL.md' files from specified local directories. 2. Boundary markers: The script relies on markdown code block delimiters to identify content for processing. 3. Capability inventory: The skill can read/write files and execute shell commands via the Graphviz utility. 4. Sanitization: The script extracts content via regex and passes it to the 'dot' command's standard input without additional validation of the Graphviz DSL content.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 12, 2026, 03:12 PM