The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill includes Python logic designed to monitor vendor service health and status pages by processing external configuration data.
Ingestion points: The VendorSLAMonitor class in SKILL.md ingests a vendors_config object containing endpoint URLs and custom headers.
Boundary markers: Absent. There are no delimiters or instructions to ignore malicious content or directives within the processed configuration data.
Capability inventory: The skill utilizes the requests library to perform network GET requests in SKILL.md.
Sanitization: Absent. The code does not validate or sanitize the target URLs provided in the configuration before execution, creating a potential surface for SSRF if the configuration is sourced from an untrusted third party.

vendor-management