The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted data from external web applications, creating a surface for potential indirect prompt injection attacks. Ingestion points: The agent visits user-provided target URLs and parses page HTML, navigation links, and console logs as described in SKILL.md. Boundary markers: No specific delimiters or isolation instructions are used to distinguish untrusted web content from the agent's core instructions. Capability inventory: The agent has the ability to interact with web pages (fill, click), create directories, and write local files (reports, screenshots, baseline JSON) as seen in the workflow sections of SKILL.md. Sanitization: The instructions include a specific rule to redact passwords from repro steps before including them in any reports.

qa