The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function involves ingesting untrusted external data—such as lead names, LinkedIn bios, and company descriptions—and interpolating them into AI-generated outbound email templates. This constitutes an inherent surface area for indirect prompt injection.
Ingestion points: External prospect data sourced from LinkedIn, Google Maps, and various CRM platforms as defined in SKILL.md and the sequence reference files.
Boundary markers: The sequence templates (e.g., references/Clay Outbound Email Prompt (3 Email Sequence).md) utilize raw placeholders like {{first_name}} and {{signal_based_observation}} without explicit delimiters or instructions to the LLM to ignore embedded commands.
Capability inventory: The skill provides instructions for designing workflows and generating copy; it does not include scripts for automated file system access, network exfiltration, or shell execution.
Sanitization: There is no evidence of data sanitization or validation protocols within the provided templates.
[CREDENTIALS_UNSAFE]: The reference materials (specifically references/Clay Personalization & Sequencing on Lemlist.md) provide guidance on configuring API keys for third-party tools like Lemlist and Octave. These instructions follow standard best practices, advising the user to manage their own keys within the legitimate platform settings rather than hardcoding or exposing them.
[EXTERNAL_DOWNLOADS]: The skill references several well-known and trusted B2B service providers (e.g., Lemlist, Apollo, Prospeo, Harmonic) and official documentation links for the Clay platform. All URLs provided are legitimate and relevant to the skill's stated purpose.

clay-gtm-outbound