security-bounty-hunter

Fail

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: CRITICALOBFUSCATIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [OBFUSCATION]: The skill contains Base64-encoded URLs within its guides, specifically designed as example payloads for XSS and data exfiltration testing.
  • Evidence found in guides/bug-bounty-report-writing-guide.md: aHR0cHM6Ly9hdHRhY2tlci5jb20vc3RlYWw/Yz0= which decodes to https://attacker.com/steal?c=.
  • [COMMAND_EXECUTION]: The skill relies heavily on the Bash tool to execute a wide array of security tools, including semgrep, sqlmap, nuclei, httpx, and subfinder. It also provides complex Bash and Python scripts within the guides for automating reconnaissance and triage pipelines.
  • [INDIRECT_PROMPT_INJECTION]: The skill possesses a broad attack surface for indirect prompt injection due to its core functionality.
  • Ingestion points: The skill actively fetches and processes untrusted external data via WebFetch, WebSearch, and command-line utilities like curl, httpx, hakrawler, and waybackurls to analyze JavaScript files and web content.
  • Boundary markers: While the skill emphasizes scope and methodology, the instructions lack explicit guidance for the agent to use delimiters or ignore embedded instructions when processing fetched data.
  • Capability inventory: The skill has access to the Bash tool (allowing arbitrary system commands) and Write/Edit tools (allowing modification of project files).
  • Sanitization: There is no documented evidence of sanitization or filtering of external content before it is processed by the agent or passed to other tools.
  • [DYNAMIC_EXECUTION]: The skill includes several Python and Bash scripts that are generated or stored as templates to be executed by the agent at runtime for automated discovery and exploitation tasks.
Recommendations
  • Contains 2 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Sep 19, 2026, 04:36 PM