The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the Bash tool to perform routine file system tasks, such as creating output directories and copying report templates, and to interact with the agent-browser CLI for web automation.
[DATA_EXPOSURE]: Authentication credentials provided by the user are handled locally to perform login actions. Session data is saved to a local file (auth-state.json) for the duration of the testing session.
[EXTERNAL_DOWNLOADS]: The skill navigates to external websites provided as the 'Target URL'. This is a core feature for web-based application testing.
[PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection by ingesting data from external websites.
Ingestion points: Web content is read via agent-browser snapshot and open commands in SKILL.md.
Boundary markers: No specific delimiters or instructions are used to separate untrusted web content from agent instructions.
Capability inventory: The skill can execute shell commands, write files, and record browser activity.
Sanitization: No filtering is applied to the DOM content retrieved from the target URL. These observations are associated with the primary purpose of web exploration and testing.

dogfood