cloudflare-analytics
Pass
Audited by Gen Agent Trust Hub on Oct 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external data via the
scan_urlandsearch_logstools, creating an attack surface for indirect prompt injection.\n - Ingestion points: External URL content and HTTP request logs are ingested into the agent context (SKILL.md).\n
- Boundary markers: The instructions lack specific delimiters or instructions to prevent the agent from obeying commands embedded in the external data.\n
- Capability inventory: The skill is limited to read-only analytics and threat intelligence; no dangerous capabilities such as file modification or shell execution are present.\n
- Sanitization: There are no documented sanitization procedures for the external content before processing.\n- [NO_CODE]: The skill does not include any scripts or executable files, reducing the risk of hidden malicious behavior.
Audit Metadata