gs-maintain-verification-skill

Pass

Audited by Gen Agent Trust Hub on Nov 19, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from the project's source code and documentation to automate testing procedures.
  • Ingestion points: The skill reads feature files and source code entry points to explain how features work (SKILL.md Step 2).
  • Boundary markers: None identified; the instructions do not specify using delimiters or warnings to ignore instructions embedded within the analyzed source code.
  • Capability inventory: The skill has the capability to execute subagents, edit project files (including .gitignore and skill directories), and run arbitrary commands via the verification skill's launch model (SKILL.md Steps 2, 4, and 5).
  • Sanitization: None identified; the skill reconciles source-derived recipes into executable actions without explicit validation or escaping of the interpreted data.
  • [COMMAND_EXECUTION]: The skill is designed to execute 'harness scripts' and a 'launch model' to drive application features during the live pass (SKILL.md Step 4). This involves running project-local executables and scripts which, while necessary for the skill's purpose, requires executing code derived from the project environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 19, 2026, 04:36 PM