create
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes data from user-supplied arguments and local recommendation files located in 'hooks/recommendations/' without performing sanitization. 1. Ingestion points: Reads user arguments via $ARGUMENTS and scans local JSON files. 2. Boundary markers: Uses tags for user input but lacks instructions to the agent to treat discovery data as untrusted. 3. Capability inventory: Invokes other skills via the Skill tool and generates installation commands for the user. 4. Sanitization: No validation or filtering is applied to the metadata fields (plugin, marketplace) before use.
- [DYNAMIC_EXECUTION]: The skill dynamically computes skill names and invokes them using the Skill tool based on strings found in local discovery files.
- [COMMAND_EXECUTION]: The skill generates shell-style commands for marketplace addition and plugin installation and instructs the user to run them based on discovery metadata.
- [EXTERNAL_DOWNLOADS]: The skill suggests the installation of external plugins from marketplaces specified in local metadata, which could lead to the execution of untrusted third-party code if the metadata source is compromised.
Audit Metadata