The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill has a high vulnerability surface for indirect prompt injection as it is designed to ingest external UI code and possesses file-modification capabilities.
Ingestion points: Audits external frontend source code and HTML templates (implied by README.md and manifest.yaml).
Boundary markers: None present; there are no clear delimiters to separate ingested code from the agent's internal reasoning.
Capability inventory: The manifest.yaml explicitly grants permissions for modifies_files and creates_artifacts.
Sanitization: No evidence of sanitization or filtering to prevent the agent from executing instructions found within audited files.
[Unverifiable Dependencies] (MEDIUM): The skill depends on external components, specifically accessibility-checker-mcp and contrast-checker-tool, without providing source verification or version locking. These external tools could potentially execute arbitrary code if compromised.

Accessibility Engineer