The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documents a pattern in examples/fragments.md that uses dangerouslySetInnerHTML to render the post.content field from a GraphQL query. This creates a vulnerability surface for Cross-Site Scripting (XSS) if the server-provided content is not properly sanitized. 1. Ingestion points: GraphQL query results in examples/fragments.md. 2. Boundary markers: Absent. 3. Capability inventory: Dynamic DOM rendering. 4. Sanitization: Absent.\n- [DATA_EXFILTRATION]: The skill utilizes localStorage for authentication token storage in examples/core.md and examples/subscriptions.md. While common in client-side tutorials, localStorage is accessible to any script on the domain, making tokens susceptible to exfiltration via XSS. Additionally, the global error link in examples/error-handling.md implements a redirect using an unvalidated redirect parameter derived from window.location.pathname, which could be exploited for Open Redirect attacks.

web-data-fetching-graphql-apollo