game-qa

Pass

Audited by Gen Agent Trust Hub on Aug 24, 2026

Risk Level: SAFECOMMAND_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill is designed to execute shell commands to launch game builds and run automated verification suites. The instructions in SKILL.md and references/qa-contract.md specify that the agent should run an 'authoritative command' to observe the game's state and capture evidence. This is a functional requirement for a QA skill and does not appear to be used for malicious purposes.
  • [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection because it processes external data sources like game design documents to influence its verification logic.
  • Ingestion points: The agent reads external project briefs and design documents (e.g., GAME_DESIGN, PRODUCT_BRIEF) to define test assertions and checkpoints as described in SKILL.md and references/test-design-method.md.
  • Boundary markers: There are no explicit instructions to use delimiters or ignore embedded instructions within the processed game design data.
  • Capability inventory: The skill has shell execution capabilities used to run game builds and verify results.
  • Sanitization: The skill does not mention specific sanitization or filtering of inputs from design documents before they are used to steer the QA process.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 24, 2026, 04:06 PM
Security Audit — agent-trust-hub — game-qa