deal-discovery-research
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill communicates with the vendor's own API (api.crawlora.net) to retrieve deal information.
- [SAFE]: The helper script (scripts/crawlora.sh) implements security best practices by using umask 077 and temporary configuration files to prevent API key exposure in process lists.
- [SAFE]: The helper script provides robust input validation, including an allowed list of API paths and sanitization of query parameters to prevent command or argument injection.
- [INDIRECT_PROMPT_INJECTION]: The skill fetches and processes community comments and deal descriptions from third-party sites like Slickdeals.
- Ingestion points: Data is ingested via the /slickdeals/comments and /slickdeals/deal endpoints in scripts/crawlora.sh.
- Boundary markers: The instructions in SKILL.md advise the agent to treat this data as "attributed observations" rather than verified facts.
- Capability inventory: The skill can execute network requests using the curl command within scripts/crawlora.sh.
- Sanitization: Query parameters are URL-encoded, but the content of external deal descriptions and comments is processed without explicit sanitization.
Audit Metadata