mlb-matchup-analyzer
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest data from several external websites including MLB.com, FanGraphs, RotoWire, and RotoBaller. This constitutes a potential ingestion surface for indirect prompt injection, as the agent processes content from these platforms to perform its analysis.\n
- Ingestion points: Data retrieved from external sports sites for probable pitcher stats, park factors, weather, and bullpen usage.\n
- Boundary markers: The skill requires the agent to cite every URL and follow a rigid output template, which helps limit the impact of any adversarial content found online.\n
- Capability inventory: The skill uses web search capabilities and file creation in the signals/ directory.\n
- Sanitization: The methodology requires all raw metrics to be normalized into a strict 0-100 scale using predefined formulas, acting as a functional filter for external data.\n- [EXTERNAL_DOWNLOADS]: The research plan relies on fetching information from well-known and authoritative sports platforms.\n
- Evidence: References to MLB.com, FanGraphs, RotoWire, and RotoBaller are utilized as primary data sources for the analyzer's inputs.\n- [NO_CODE]: The skill consists entirely of markdown instructions, methodology guides, and JSON rubrics; it contains no executable scripts, binaries, or dynamic context injection patterns.
Audit Metadata