The Agent Skills Directory

[Prompt Injection] (SAFE): No instructions attempting to override agent behavior or bypass safety filters were found. The instructional content is limited to software development patterns.
[Data Exposure & Exfiltration] (SAFE): No hardcoded credentials, sensitive file path leaks, or unauthorized network calls were detected. The skill explicitly provides patterns for secure data handling and authentication.
[Remote Code Execution] (SAFE): While the skill's metadata allows for npm and npx command execution, there are no patterns involving the download and execution of untrusted remote scripts (e.g., curl | bash).
[Indirect Prompt Injection] (LOW):
Ingestion points: The skill interacts with the local codebase (React components, API routes) through Read, Write, and Edit tools.
Boundary markers: None explicitly defined in the provided markdown.
Capability inventory: Includes file system modification (Write, Edit) and package management commands (npm, npx).
Sanitization: The skill contains explicit instructions to sanitize error logs and aggregate data to maintain HIPAA compliance, reducing the risk of accidental data leakage.

admin-dashboard