The Agent Skills Directory

[COMMAND_EXECUTION]: The skill performs REST API operations including POST, PATCH, and DELETE against 'api.vapi.ai' to manage lifecycle resources like personalities, scenarios, and suites. This allows for active mutation of remote state on the Vapi platform.
[DATA_EXPOSURE_AND_EXFILTRATION]: Configuration payloads and simulation parameters are transmitted to 'api.vapi.ai'. Since 'vapiai' is the identified vendor, these operations are classified as legitimate vendor-resource interactions and do not constitute unauthorized exfiltration.
[INDIRECT_PROMPT_INJECTION]: The skill processes conversation transcripts and results from the 'api.vapi.ai' endpoint which could contain adversarial content. 1. Ingestion points: Data is retrieved from the '/eval/simulation/run/{id}/item' endpoint. 2. Boundary markers: No specific delimiters are used to isolate transcript content from agent instructions during analysis. 3. Capability inventory: The skill has permissions to mutate and delete resources via the Vapi REST API. 4. Sanitization: No explicit content filtering or escaping of transcript data is implemented before the agent performs its diagnostic analysis.

simulations