skills/goldk3y/skills/wizard-quest/Gen Agent Trust Hub

wizard-quest

Pass

Audited by Gen Agent Trust Hub on Aug 4, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a bundled local shell script, scripts/consult_the_fates.sh, to track quest history and generate randomized research themes. The script uses standard utilities like awk and date with appropriate input handling.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface because it is instructed to read and follow 'dangling threads' from markdown files within the repository's scrolls/ directory. An attacker who can commit files to the repository could theoretically influence the agent's research direction or web search parameters.
  • Ingestion points: Markdown files located in the scrolls/ directory at the repository root.
  • Boundary markers: The 'Wizard's Oath' in SKILL.md acts as a strong set of negative constraints, specifically 'Oath 5' which prohibits the agent from modifying any files except the new research scroll.
  • Capability inventory: The skill has access to web search, local file reading/writing, and execution of its internal shell script.
  • Sanitization: The consult_the_fates.sh script performs structural parsing of file frontmatter using awk rather than interpreting the files as executable content.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 4, 2026, 10:34 PM
Security Audit — agent-trust-hub — wizard-quest