The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is inherently designed to ingest and process untrusted content from external sources, which creates a surface for indirect prompt injection.
Ingestion points: The skill explicitly instructs the agent to read raw research materials such as customer interview transcripts, survey results, support tickets, and reviews. It also provides detailed playbooks in references/source-guides.md for mining content from public platforms like Reddit, G2, Hacker News, and LinkedIn.
Boundary markers: The instructions do not explicitly define boundary markers (e.g., delimiters) to separate untrusted data from the agent's internal instructions, nor do they include specific warnings to ignore instructions embedded within the research data.
Capability inventory: The skill is capable of generating research reports, personas, and VOC (Voice of Customer) banks. It also references hand-offs to other skills such as copywriting and page-cro where synthesized data might be used in downstream tasks.
Sanitization: No sanitization, filtering, or validation mechanisms for external content are mentioned in the instructions.

customer-research