The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/with_server.py utility is designed to execute arbitrary shell commands. It utilizes subprocess.Popen with shell=True to launch server processes and subprocess.run to execute the specified automation commands. The SKILL.md instructions explicitly tell the agent to use these scripts as 'black boxes' and avoid reading the source code before execution, which discourages the agent from verifying the safety of the shell commands it generates.\n- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of data from external web applications and browser logs, creating an attack surface for indirect prompt injection.\n
Ingestion points: The agent captures DOM content, element attributes, and console messages as shown in examples/element_discovery.py and examples/console_logging.py.\n
Boundary markers: There are no protective delimiters or instructions to ignore embedded commands within the ingested data.\n
Capability inventory: The skill environment grants the agent capabilities for file system writes and arbitrary command execution.\n
Sanitization: The skill performs no sanitization or validation on content retrieved from the web browser before it is added to the agent's context.

webapp-testing