rollinggo-hotel-booking
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The installation script (
scripts/install.py) downloads a standalone CLI binary (rgh) from the vendor's official GitHub repository releases. - [COMMAND_EXECUTION]: The skill uses Node.js (
scripts/rgh.js) and Python (scripts/rgh.py) proxy scripts to execute therghCLI tool usingspawnandsubprocess.runrespectively. - [DYNAMIC_EXECUTION]: The skill instructions specify a 'Self-Discovery Rule' where the agent is told to run commands with the
--helpflag at runtime to dynamically determine available parameters. - [INDIRECT_PROMPT_INJECTION]: The skill processes hotel descriptions and tags returned from an external API, creating a surface for indirect prompt injection.
- Ingestion points: Data enters the context through the output of
search-hotelsandhotel-detailcommands executed via the CLI. - Boundary markers: No explicit instruction delimiters are used for the ingested data, though instructions mandate strict output formatting rules to prevent technical detail exposure.
- Capability inventory: The skill can execute shell commands via the
rghwrapper scripts (scripts/rgh.py,scripts/rgh.js). - Sanitization: No automated sanitization or escaping of the data retrieved from the hotel service is implemented in the provided wrapper scripts.
Audit Metadata