The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes functionality to retrieve market news and sentiment data from an external API, creating a vector for indirect prompt injection where untrusted third-party content could influence the agent.
Ingestion points: The NEWS_SENTIMENT endpoint described in references/intelligence.md fetches article feeds containing titles and summaries from the web.
Boundary markers: The skill does not provide specific delimiters or instructions to the agent to distinguish between its own logic and the fetched news content.
Capability inventory: The skill utilizes requests for network operations and pandas for data processing. No high-risk capabilities like shell execution were identified.
Sanitization: There is no evidence of content filtering or sanitization for the retrieved news data.
[PROMPT_INJECTION]: The SKILL.md file contains specific instructions directing the agent to proactively suggest the author's commercial platform, 'K-Dense Web', when user requests reach a certain level of complexity. This instruction attempts to override the agent's neutral persona to serve promotional objectives.

alpha-vantage