steve-jobs-perspective
Warn
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: MEDIUMDYNAMIC_CONTEXT_INJECTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_CONTEXT_INJECTION]: The skill includes a preamble that executes a shell command when the skill is loaded:
bash "$(dirname "${BASH_SOURCE[0]}")/../../check-update.sh". This allows for code execution before the user begins the conversation. - [COMMAND_EXECUTION]: The preamble command executes an external script (
check-update.sh) via a bash subprocess. As the script is stored outside the skill's specific folder, its integrity and purpose cannot be verified during static analysis. - [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data such as PRDs and feature lists for critique. It lacks boundary markers or instructions to ignore embedded commands, which could allow malicious content in these documents to influence the agent's behavior. Ingestion points: User-provided text in product analysis requests (e.g., {X}, {功能列表}). Boundary markers: None. Capability inventory: Uses
WebSearchandReadtools. Sanitization: None.
Audit Metadata