The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill manages interactions with external data via provisioning webhooks, which presents a potential injection surface. The author proactively mitigates this by instructing the agent to treat all inbound payloads as untrusted and to validate them before use. * Ingestion points: Webhook payload data referenced in the agent instructions. * Boundary markers: Explicit security guidelines are provided to avoid interpolation of external data into prompts or commands. * Capability inventory: The skill is limited to authenticated network requests (curl) to official endpoints. * Sanitization: Proactive instructions for data validation and sanitization are included in the security section.
[EXTERNAL_DOWNLOADS]: The skill retrieves documentation and API specifications from official vendor domains (developers.sinch.com), which is an expected and safe operation for an API-focused skill.

sinch-provisioning-api