The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill possesses an attack surface for indirect prompt injection because it ingests and processes untrusted HTML data provided by the user.
Ingestion points: scripts/analyze.sh reads and processes the content of user-provided HTML files using grep.
Boundary markers: Absent. There are no instructions to the agent to treat the content of the analyzed files as data rather than instructions.
Capability inventory: The script is limited to local file system read operations and pattern matching. It does not have network access or write capabilities.
Sanitization: No sanitization or escaping of the HTML content is performed before the agent reviews it.
[Command Execution] (SAFE): The bash script scripts/analyze.sh uses standard utilities (grep, find) and properly quotes variables to prevent shell injection via malicious filenames.

web-quality-audit