trigger-dev-tasks

Fail

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: CRITICALEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • EXTERNAL_DOWNLOADS (MEDIUM): The skill guides the agent on using 'aptGet' and 'additionalPackages' extensions in 'config.md'. These functions facilitate the installation of arbitrary system packages and NPM packages, which introduces a dependency risk.
  • REMOTE_CODE_EXECUTION (MEDIUM): Documentation in 'config.md' for 'pythonExtension' describes methods like 'python.runInline' and 'python.runScript', enabling the execution of Python code within the task environment.
  • COMMAND_EXECUTION (LOW): The skill allows the use of the 'Bash' tool and demonstrates build-time commands such as 'chmod +x' in 'config.md'. While appropriate for containerized tasks, these are sensitive operations.
  • PROMPT_INJECTION (LOW): The skill is susceptible to indirect prompt injection due to its ingestion of user-controlled code files while possessing high-capability tools. 1. Ingestion points: 'Read', 'Glob', and 'Grep' tools in 'SKILL.md'. 2. Boundary markers: Absent. 3. Capability inventory: 'Bash', 'Write', and Python execution. 4. Sanitization: Absent.
  • Note on Automated Scan (SAFE): The automated alert regarding 'logger.info' is a false positive where a common code method was misidentified as a blacklisted URL.
Recommendations
  • Contains 1 malicious URL(s) - DO NOT USE
Audit Metadata
Risk Level
CRITICAL
Analyzed
Feb 17, 2026, 06:31 PM