fallback-and-resilience
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill references an external Docker image hosted on the GitHub Container Registry (
ghcr.io/borski/sw-fares) as a fallback tool for fetching Southwest airline fares. This resource is hosted under the skill author's own namespace. - [COMMAND_EXECUTION]: The instructions include diagnostic and maintenance commands for the agent to use or suggest to the user, such as
curl ifconfig.meto determine public IP addresses for whitelisting and variousdockercommands (logout,pull,build) for managing the skill's environment. - [INDIRECT_PROMPT_INJECTION]: The skill defines a large attack surface by design, as it instructs the agent to process and fallback between numerous external data sources including travel APIs (Duffel, Kiwi.com, Seats.aero), web scraping targets (Airbnb, Southwest), and community-driven content (Wikipedia).
- Ingestion points: Multiple external API responses and scraped web content from sources listed in the fallback table in
SKILL.md. - Boundary markers: The documentation does not provide specific instructions for using delimiters or boundary markers when interpolating this external data into the agent's context.
- Capability inventory: The skill guides an agent that possesses network access capabilities, browser-based scraping tools (Patchright), and local command execution for Docker management.
- Sanitization: No explicit sanitization or filtering logic is defined for the external inputs mentioned.
Audit Metadata