The Agent Skills Directory

[COMMAND_EXECUTION]: The helper utility scripts/with_server.py uses subprocess.Popen with shell=True to execute server commands and subprocess.run to execute the automation task. This implementation allows for the execution of arbitrary shell strings provided via the --server and positional command arguments.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a vulnerability surface for indirect prompt injection.
Ingestion points: The agent is instructed to navigate to local web applications and inspect rendered DOM content (page.content(), page.screenshot()) in SKILL.md and examples/element_discovery.py.
Boundary markers: None. There are no instructions to ignore or delimit instructions found within the web applications being tested.
Capability inventory: The skill provides full shell execution capabilities through the scripts/with_server.py script, which can be invoked by the agent to start servers or run tests.
Sanitization: None. Data extracted from web pages that might influence command arguments is not sanitized before being passed to the shell-executing helper script.
[COMMAND_EXECUTION]: In SKILL.md, the author includes a 'Best Practice' advising the agent to use bundled scripts as 'black boxes' and explicitly tells the agent 'DO NOT read the source until you try running the script first'. While framed as a context window optimization, this instruction discourages the agent from performing a security review of the script's shell execution behavior before invoking it.

webapp-testing