skills/mukul975/anthropic-cybersecurity-skills/performing-cloud-log-forensics-with-athena/Gen Agent Trust Hub
performing-cloud-log-forensics-with-athena
Pass
Audited by Gen Agent Trust Hub on Apr 11, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [PROMPT_INJECTION]: The skill processes untrusted external log data, creating a surface for indirect prompt injection. * Ingestion points: AWS log sources (CloudTrail, VPC Flow, ALB, S3) accessed via
scripts/agent.py. * Boundary markers: No delimiters or ignore-instructions warnings are used in the resulting JSON report. * Capability inventory: The skill utilizesboto3to perform Athena queries and access S3 objects. * Sanitization: Log data is extracted and summarized without content validation or escaping. - [COMMAND_EXECUTION]: The
scripts/agent.pyscript constructs and executes SQL queries against the AWS Athena service based on user-provided parameters like dates and database names. - [DATA_EXFILTRATION]: The skill is designed to read and consolidate sensitive cloud audit logs. Access is managed through the user's local AWS credentials, and output is saved to user-specified local files or S3 buckets.
Audit Metadata