skills/aliyun/alibabacloud-aiops-skills/alibabacloud-web-application-attacks-analysis/Gen Agent Trust Hub
alibabacloud-web-application-attacks-analysis
Warn
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: MEDIUMPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The instructions in
SKILL.mdcontain an 'ABSOLUTE PROHIBITION' explicitly forbidding the agent from reading, grepping, or browsing the analyzer's source code (scripts/log_analyzer.py) before execution. This pattern is used to manipulate the agent into bypassing security inspection of local executable code. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (web access logs) to produce structured findings designed to influence the behavior of downstream agents.
- Ingestion points: The
log_fileparameter inscripts/log_analyzer.py, which reads and parses external log records. - Boundary markers: The skill lacks explicit instructions or markers to isolate the parsed log data or warn the agent to ignore instructions embedded within the processed logs.
- Capability inventory: The skill uses
scripts/log_analyzer.pyto read files and generate both human-readable and machine-readable (JSON) reports. The output JSON is specifically intended as input for subsequent agent stages. - Sanitization: While the script uses regular expressions and standard URL parsing, it does not perform robust sanitization or escaping of log content to prevent the injection of malicious instructions into the reports.
Audit Metadata