sports-scores-research
Pass
Audited by Gen Agent Trust Hub on Aug 14, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external third-party sports data sources (ESPN, SofaScore, and MLB) through the Crawlora API. This creates a surface for indirect prompt injection, where malicious instructions hidden in external data (such as player biographies, news headlines, or team descriptions) could influence the agent's behavior.
- Ingestion points: The script
scripts/crawlora.shfetches data from theapi.crawlora.netbase URL, specifically for endpoints defined inreference/endpoints.mdsuch as/espn/athlete,/espn/news, and/sofascore/search. - Boundary markers: Absent. The instructions do not define clear delimiters or provided safety prompts to the agent to treat the API response content as untrusted data.
- Capability inventory: The agent uses this data to answer user queries; it does not have high-privilege capabilities such as arbitrary shell execution or local file writing that would be triggered by the API content.
- Sanitization: Absent. The skill relies on the agent's internal logic to parse and present the data without explicit sanitization of the external strings.
Audit Metadata