The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to analyze Amazon marketplace data, including user-generated product reviews, to generate "selling point" and "pain point" summaries. This content is untrusted and could contain malicious instructions designed to manipulate the agent's logic or conclusions.
Ingestion points: Data returned by tools such as asin_detail and review (found in SKILL.md) contains raw text from the Amazon website.
Boundary markers: The instructions do not specify any delimiters or special handling to isolate the third-party review data from the agent's internal reasoning.
Capability inventory: The skill possesses network communication capabilities and the ability to read/write local cached files, which could be leveraged if an injection occurs.
Sanitization: There are no instructions for sanitizing or filtering the content ingested from the external data source.
[EXTERNAL_DOWNLOADS]: The skill requires network access to a third-party API hosted by the vendor to perform its tasks.
URLs: The skill communicates with https://www.iqinghu.com/api/desktop/qh/mcp as its main data endpoint.
Mechanism: It uses standard HTTP methods (POST) to invoke remote tools using a JSON-RPC format.
[COMMAND_EXECUTION]: The skill instructions explicitly direct the agent to use command-line or library tools to perform network requests.
Evidence: The documentation suggests the use of curl, fetch, or requests for interacting with the Qinghu data interface, implying that the agent has the capability to execute environment-level commands or scripts.

qinghu-amazon-asin-analyst