The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external websites, such as competitor news and landing pages. This creates a surface for indirect prompt injection where malicious instructions could be embedded in the content being monitored. The skill mitigates this risk by enforcing strict formatting that separates factual data from AI-generated interpretation, limiting quotes to a maximum of 140 characters, and requiring date verification from the source content to prevent stale or manipulated data from being processed as current.\n- [COMMAND_EXECUTION]: The skill uses the platform's native command-line utility (clay whoami) to verify workspace status and tool availability. This is a standard environmental check used to ensure the skill can function correctly within its intended platform context.\n- [DATA_EXFILTRATION]: The skill performs network-based scraping of competitor websites to gather intelligence. While it involves external network access, the skill includes explicit instructions forbidding the writing of data to external CRMs and specifically warns against saving sensitive credentials or tokens when persisting user configuration.

competitive-intelligence-radar