The Agent Skills Directory

[SAFE]: The skill demonstrates secure practices, such as managing sensitive keys via environment variables and implementing authenticated error reporting.
[DATA_EXFILTRATION]: The instructions include patterns for reporting logs and errors to external endpoints, including the well-known service Sentry and user-defined logging endpoints. These are standard observability features and do not represent unauthorized exfiltration.
[INDIRECT_PROMPT_INJECTION]: The skill defines middleware that processes external event data (ctx.event.data).
Ingestion points: Event data is ingested in onFunctionRun hooks across several examples in references/built-in-middleware.md and references/dependency-injection.md.
Boundary markers: None explicitly provided for raw data separation.
Capability inventory: Uses fetch for error reporting and Sentry for tracing.
Sanitization: No specific content sanitization is demonstrated before data is logged or sent to reporting endpoints.
[EXTERNAL_DOWNLOADS]: The skill references several well-known and official packages from the Inngest ecosystem and other established providers (OpenAI, Stripe, Prisma, Sentry), which is expected for the stated purpose of the skill.

inngest-middleware