The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies raw sales call transcripts as its primary input, which are treated as untrusted data (Ingestion points). It lacks explicit boundary markers, such as XML tags or specific delimiters, to separate the transcript content from the skill's instructions (Boundary markers: absent). The analysis output is explicitly designed to guide downstream actions in other skills like follow-up-emails and multithread-outreach (Capability inventory). No instructions for sanitizing or filtering the provided transcripts are included in the requirements (Sanitization: absent).

call-analysis