The Agent Skills Directory

[DATA_EXFILTRATION]: The skill accesses sensitive personal and professional data to perform its core functions.
Evidence: The skill reads search/context/experience-library.yaml (STAR stories, accomplishments), search/context/career-plan.yaml (target levels, weaknesses), and search/context/interview-history.yaml (past patterns and weak areas).
Evidence: The presence of network-enabled tools like WebSearch and WebFetch creates a potential data exposure surface, though no explicit exfiltration logic is present in the current script.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its reliance on external data sources.
Ingestion points: The skill ingests untrusted data from the web via WebSearch and WebFetch while researching company culture and interview questions in Step 2.
Boundary markers: Absent. The skill does not use specific delimiters or instructions to the model to ignore potential injection attempts within the fetched web content.
Capability inventory: The skill possesses Read, Write, and network access tools, which could be exploited if the model follows malicious instructions embedded in web search results.
Sanitization: Absent. There is no evidence of validation or filtering of the content retrieved from external websites before it is used to generate the interview prep package.

interview-prep