The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from Naver blog titles and keyword autocompletes, which is then used to generate a writing brief (draftPrompt) passed to other agents.
Ingestion points: Data enters the context via diagnose_naver_blog (reading public blog titles and categories) and expand_naver_keywords (fetching autocomplete results from Naver).
Boundary markers: The skill instructions do not specify the use of delimiters (e.g., XML tags or triple quotes) to isolate external content, nor does it instruct the agent to ignore instructions embedded within blog titles.
Capability inventory: The skill has the capability to scan blog visibility, research metrics, and hand off generated prompts to the naver-post and naver-ai-cite skills for content creation.
Sanitization: There is no mention of sanitizing, escaping, or filtering the retrieved blog titles or autocompletes before they are interpolated into the resulting prompt.

naver-keywords