The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill accepts untrusted data through the $ARGUMENTS variable, including test file paths and descriptions, which are interpolated into shell commands. This creates a surface for indirect prompt injection or command injection.\n
Ingestion points: The $ARGUMENTS object in SKILL.md receives external data.\n
Boundary markers: No delimiters or instructions to ignore embedded commands are present in the prompt templates.\n
Capability inventory: The skill performs shell execution using npx playwright test across multiple steps in SKILL.md.\n
Sanitization: There is no evidence of input validation, path sanitization, or escaping of the <file> argument before it is passed to the shell.\n- [COMMAND_EXECUTION]: The skill instructions direct the agent to execute shell commands using npx to run, repeat, and trace Playwright tests. This behavior is consistent with the skill's stated purpose of debugging test failures.\n- [EXTERNAL_DOWNLOADS]: The skill uses npx playwright, which may involve downloading the Playwright package or browser binaries from well-known registries if they are not already present in the environment.

fix