web-sqli
Fail
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: CRITICALINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill inherently interacts with untrusted external data through the
WebFetchandWebSearchtools, as well as via custom scanning scripts likeguides/bug_bounty_sqli_scanner.pythat crawl target websites. - Ingestion points: Data enters the agent context through the analysis of HTTP responses, forms, and page content during vulnerability scanning.
- Boundary markers: Absent. The instructions do not provide explicit guidance or delimiters to prevent the agent from accidentally obeying instructions embedded in the target site's content (e.g., hidden HTML comments).
- Capability inventory: The agent has access to
Bash(for running tools like sqlmap) andWrite(for saving scan reports). - Sanitization: Absent. The scripts and instructions do not specify methods for sanitizing external content before it is processed by the agent.
- [COMMAND_EXECUTION]: The skill instructions and automation scripts (e.g.,
guides/sqli-labs-auto.sh) utilize shell commands viacurlandsqlmapto perform security testing. While these are part of the skill's primary purpose, they involve the execution of subprocesses and network operations. - [EXTERNAL_DOWNLOADS]: The skill references several well-known security education and tool resources, including
portswigger.net,owasp.org,sqlmap.org, andhacktricks.xyz. These are documented neutrally as established services within the security research community.
Recommendations
- Contains 2 malicious URL(s) - DO NOT USE
Audit Metadata