ios-design-review
Pass
Audited by Gen Agent Trust Hub on Nov 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple local binaries within the
~/.claude/skills/gstack/bin/directory (e.g.,gstack-skill-start,gstack-question-log,gstack-skill-end). These are utility scripts provided by the author's framework for managing session initialization, preference logging, and telemetry. - [INDIRECT_PROMPT_INJECTION]: The skill audits the visual design of third-party iOS applications by retrieving screenshots and accessibility tree metadata (
GET /elements), creating an untrusted data ingestion surface. - Ingestion points: UI element text and visual data are ingested from the target application via the
gstack-ios-qa-daemonduring the analysis loop. - Boundary markers: The agent is directed to follow a specific 10-dimension rubric (typography, spacing, accessibility, etc.) which provides a structured scope for the analysis.
- Capability inventory: The skill uses
Bashto perform network requests to the local daemon and write findings to markdown reports in the local filesystem. - Sanitization: There are no explicit instructions for sanitizing or escaping text content extracted from the mobile application before it is processed by the agent.
Audit Metadata