The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill documentation describes a template rendering system that processes external data, creating a potential surface for indirect prompt injection.
Ingestion points: Data enters the agent context via the TemplateModel and TestRenderModel objects described in SKILL.md and references/template-api.md.
Boundary markers: The documentation explicitly differentiates between double braces {{variable}} (which perform HTML escaping) and triple braces {{{variable}}} (which disable escaping for raw HTML injection).
Capability inventory: The skill provides instructions for sending emails via client.sendEmailWithTemplate and client.sendEmailBatchWithTemplates, which execute network requests to the Postmark API.
Sanitization: While Handlebars provides default escaping, the skill documents the use of triple braces for unescaped content, which requires the developer to ensure the source data is trusted.
[SAFE]: The skill uses official Postmark API endpoints (api.postmarkapp.com) and requires the official postmark Node.js library, which are well-known and trusted vendor resources.
[SAFE]: The skill follows security best practices for credential management by demonstrating the use of environment variables (process.env.POSTMARK_SERVER_TOKEN) rather than hardcoding sensitive keys.

postmark-templates