The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user responses and a free-text field, which are then written into a local markdown file. This creates an indirect prompt injection surface should the agent or a user process this file later. However, this is part of the core functionality for generating a consultation brief. * Ingestion points: Responses to the AskUserQuestion rounds and the final free-text additional details field. * Boundary markers: No explicit delimiters or instruction-isolation markers are used in the markdown template. * Capability inventory: Uses file-writing capabilities to save the brief to the local directory. * Sanitization: Input is captured and saved verbatim without explicit sanitization.
[NO_CODE]: The skill is composed strictly of natural language instructions and does not contain any embedded scripts, executables, or third-party dependencies.

site-diagnosis