kill-ai-slop
Warn
Audited by Gen Agent Trust Hub on Oct 5, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [DYNAMIC_EXECUTION]: The scanner script
scripts/scan.mjsallows loading external JavaScript modules via the--rulescommand-line argument using the dynamicimport()function. This creates a risk of arbitrary code execution if a malicious rules file is provided or if the agent is manipulated into loading an untrusted file from the project.\n- [INDIRECT_PROMPT_INJECTION]: The skill's workflow involves reading untrusted source code from the project and presenting it to the agent for triage, creating a vulnerability surface for indirect prompt injection.\n - Ingestion points: The
scripts/scan.mjsscript reads various frontend source files (HTML, CSS, JS, etc.) across the project directory.\n - Boundary markers: There are no explicit delimiters or instructions to help the agent distinguish between its instructions and the content of the files being scanned.\n
- Capability inventory: The agent is instructed in
SKILL.mdto modify the codebase to apply fixes, providing a vector for state changes based on injected content.\n - Sanitization: No sanitization or escaping is performed on the code snippets before they are analyzed by the agent.
Audit Metadata