The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references installation guides and related components hosted on the author's official GitHub repository (github.com/inference-sh/skills).
[PROMPT_INJECTION]: The skill processes untrusted web data, creating a surface for indirect prompt injection where malicious instructions on a webpage could influence agent behavior.
Ingestion points: Untrusted content is ingested into the agent context from the web via tavily/extract and exa/extract apps as seen in SKILL.md.
Boundary markers: The provided usage examples do not implement delimiters or isolation techniques to separate extracted web content from model instructions.
Capability inventory: The skill allows execution of various infsh apps which can perform data retrieval and aggregation across the platform.
Sanitization: No explicit sanitization or filtering of the fetched web content is described in the skill's workflows.

web-search