The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were detected in this skill.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process retrieved context from external sources. * Ingestion points: The chunks parameter used in formatting and budget functions in SKILL.md. * Boundary markers: Explicitly recommends using structured XML tags like <source> to delimit context and provides system prompt instructions to enforce those boundaries. * Capability inventory: Logic is limited to formatting, budget calculation, and sorting. No capabilities for file system modification or unauthorized network access are present. * Sanitization: Recommends server-side validation of citations to detect hallucinated source IDs.
[EXTERNAL_DOWNLOADS]: The skill includes an example reference to Stripe's official documentation at https://stripe.com/docs/webhooks. This targets a well-known service for educational purposes and does not represent a security risk.

context-injection