cloud-security-scoutsuite
Fail
Audited by Gen Agent Trust Hub on Feb 14, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS] (MEDIUM): The skill requires the installation of the 'scoutsuite' Python package. This is an unversioned external dependency from a non-trusted source (NCC Group), which could lead to supply chain risks or unexpected behavior changes.
- [COMMAND_EXECUTION] (MEDIUM): The skill provides instructions to execute the 'scout' command with various arguments to perform cloud scans. This grants the agent the capability to invoke external processes that interact directly with the cloud provider's API.
- [DATA_EXFILTRATION] (HIGH): The skill directs the agent to 'Read JSON output and present findings'. This exposes highly sensitive cloud security posture data, including misconfigurations and vulnerability details, to the agent's context. If the agent is connected to untrusted channels, this data could be leaked.
- [PROMPT_INJECTION] (HIGH): The skill is vulnerable to Indirect Prompt Injection (Category 8). It ingests untrusted data from external sources (cloud resource names, tags, and configurations) and instructs the agent to process and summarize it. An attacker with control over cloud resource metadata could embed malicious instructions that override the agent's behavior during the reporting phase. There are no boundary markers or sanitization steps provided in the instructions.
Recommendations
- AI detected serious security threats
Audit Metadata