android-dev

Pass

Audited by Gen Agent Trust Hub on Apr 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses bundled shell scripts (scaffold-feature.sh and scaffold-usecase.sh) to automate the creation of project modules and files. The agent is instructed to run these scripts as part of its primary functionality.
  • [PROMPT_INJECTION]: The skill contains an attack surface for indirect prompt injection in its scaffolding logic: 1. Ingestion points: User-provided feature names, package names, and repository names in SKILL.md. 2. Boundary markers: Absent; the scripts do not use delimiters or instructions to ignore instructions embedded in the user data. 3. Capability inventory: The agent has the capability to execute the bundled shell scripts which perform directory creation (mkdir) and file writing (cat >) to the local project structure. 4. Sanitization: Absent; the scripts do not validate or escape user arguments, which could allow for path traversal (e.g., using ../ in a feature name) or code injection into the generated Kotlin and Gradle build files via crafted package strings.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 17, 2026, 12:03 AM