The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill provides integration code templates that rely on environment variables for sensitive API credentials. While it correctly suggests using .env files, it explicitly identifies keys like NAMECHEAP_API_KEY, GODADDY_API_SECRET, and DOMAIN_API_KEY in the documentation, which could be targeted by other malicious skills.
[INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to 'Scan Project Documentation' and 'Review validation libraries in use' before providing guidance. This process ingests untrusted external data that could contain malicious instructions designed to influence the agent's output or validation logic.
Ingestion points: Project documentation files, configuration files, and validation libraries scanned during context discovery (SKILL.md).
Boundary markers: None specified; the skill lacks explicit instructions to ignore embedded prompts in scanned project files.
Capability inventory: No direct shell execution or file-write capabilities are present in the skill's own code; it primarily generates implementation guidance.
Sanitization: No sanitization or filtering logic is provided for the data gathered during the project context discovery phase.
[DATA_EXFILTRATION]: The skill provides templates for making external HTTP requests to third-party domain registrar APIs (Namecheap, GoDaddy). While these are intended for domain availability checks, a compromised or malicious agent could repurpose these network patterns to exfiltrate sensitive data retrieved during the context discovery phase to these or other external endpoints.

search-domain-validator