ctf-flag-hunting

Pass

Audited by Gen Agent Trust Hub on Sep 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides numerous command-line examples for system reconnaissance and file searching, including find, grep, cat, env, ps, netstat, and lsof to be executed on a target host environment.\n- [DATA_EXFILTRATION]: The skill provides specific instructions for identifying and accessing sensitive files on a target system, such as /etc/passwd, environment configuration files (.env), and database connection scripts (config.php, app.py). It also discusses techniques for blind data extraction, such as DNS out-of-band (OOB) exfiltration.\n- [INDIRECT_PROMPT_INJECTION]: The skill involves reading external files (via LFI) and database contents (via SQLi) from target systems, which constitutes a potential ingestion surface for untrusted data that could influence the agent.\n
  • Ingestion points: Target file content retrieval in Phase 3 and database query results in Phase 4.\n
  • Boundary markers: None provided to distinguish between target-sourced data and agent instructions.\n
  • Capability inventory: Shell command execution, file system access, and network utility usage.\n
  • Sanitization: No specific mechanisms for sanitizing or validating the retrieved external content are described.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 16, 2026, 07:51 PM
Security Audit — agent-trust-hub — ctf-flag-hunting