The Agent Skills Directory

[PROMPT_INJECTION]: The skill architecture involves processing and displaying external data (user metrics, access requests, and activity logs) within an administrative dashboard. This creates a surface for indirect prompt injection where malicious content embedded in the ingested data could potentially influence the agent's behavior during analysis or development tasks.
Ingestion points: The dashboard fetches data from multiple internal API endpoints such as /api/admin/stats and /api/admin/access-requests as defined in src/app/admin/page.tsx.
Boundary markers: The provided code snippets do not include explicit instructions or delimiters (such as XML tags or "ignore instructions" warnings) to separate external data from the agent's primary instructions.
Capability inventory: The skill allows high-privilege tools including Bash(npm:*, npx:*), Write, and Edit, which could be exploited if the agent is misled by injected content.
Sanitization: While the skill demonstrates robust access control (requireAdmin) and audit logging (logAdminAction), it does not explicitly detail the sanitization or escaping of external content before it is interpolated into the UI or processed by the agent.

admin-dashboard