The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external prospect data to generate personalized email sequences and a local HTML QA tool, creating an injection surface.
Ingestion points: Enriched prospect lists in *-leads.csv format are ingested from the prospect-scraper skill or other external sources, as described in SKILL.md.
Boundary markers: The skill uses {{token}} interpolation and spintax {a|b|c} to separate user-provided data from instruction templates.
Capability inventory: The skill generates self-contained HTML files (* Email Preview.html) for review and interacts with the Smartlead API to send outbound communications.
Sanitization: While the documentation in SKILL.md and references/preview-tool-standard.md mentions highlighting missing data, it does not explicitly specify HTML escaping or sanitization for rendering untrusted prospect data in the generated preview tool.
[COMMAND_EXECUTION]: The skill instructions require the agent to execute specific command-line utilities to perform infrastructure checks and data management.
DNS Verification: SKILL.md instructs the agent to use the dig utility to verify SPF, DKIM, DMARC, and MX records for sending domains.
API Interaction: The instructions suggest using a Smartlead CLI or API wrapper to perform bulk operations on lead data.

outbound-engine