The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it is designed to ingest and analyze untrusted data in the form of 'candidate_content' (generated product descriptions).
Ingestion points: The skill processes content from the generated_content table and candidate_content fields.
Boundary markers: The provided instructions do not specify any delimiters or boundary markers to separate the rubric instructions from the content being evaluated.
Capability inventory: The skill itself is instructional and used for scoring/evaluation; however, if integrated into an automated pipeline, malicious instructions within the product descriptions could potentially influence the evaluation outcome.
Sanitization: There is no evidence of sanitization or filtering of the input content before it is processed by the agent using this rubric.

quality-evaluation