The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/with_server.py uses subprocess.Popen with shell=True to execute server startup commands provided via the --server argument. This is used to support complex startup commands including directory changes and chained operations.
[COMMAND_EXECUTION]: The script scripts/with_server.py utilizes subprocess.run to execute automation commands passed as arguments after the managed servers have started.
[PROMPT_INJECTION]: The SKILL.md file instructs the agent to treat scripts as "black-box scripts" and discourages reading the source code to save context window space. While practically motivated for LLMs, this can theoretically be used to reduce oversight of the scripts' internal operations.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to its core functionality.
Ingestion points: The scripts in examples/element_discovery.py and examples/console_logging.py read DOM content and console output from external, potentially untrusted web applications.
Boundary markers: None identified in the provided scripts or instructions.
Capability inventory: The skill provides capabilities for shell command execution (scripts/with_server.py), file system access via screenshots, and local file reading (examples/static_html_automation.py).
Sanitization: No sanitization of ingested web content is performed before processing.

webapp-testing