The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes unstructured content from external web pages, which could potentially contain malicious instructions intended to influence the agent's behavior during summarization.
Ingestion points: scripts/web_search.py fetches web content via a proxy API and returns it to the agent.
Boundary markers: The SKILL.md documentation provides instructions on how to handle the data ("summarize directly") but does not include explicit delimiters or instructions to ignore embedded commands within the retrieved text.
Capability inventory: The skill is capable of network operations (to the search API) and local file writing (to the nexscope directory). It does not have capabilities for subprocess execution, code evaluation, or privilege escalation.
Sanitization: No explicit sanitization or filtering of the retrieved web content is performed before it is presented to the agent for summarization.
[DATA_EXPOSURE]: The script scripts/web_search.py automatically writes search results and API responses to the local filesystem under a nexscope directory located in either the current working directory, the user's home directory, or a temporary folder. This is used for session-based logging and caching of search data.

ecommerce-web-search