The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill contains a vulnerability surface where untrusted data from user arguments is processed and used in exploitable capabilities (shell execution).
Ingestion points: The $ARGUMENTS field in SKILL.md captures a test file path and test name directly from user input.
Boundary markers: The skill lacks boundary markers or explicit instructions to the agent to treat the input as data only and ignore any embedded shell control characters.
Capability inventory: The skill utilizes npx playwright test shell commands (found in SKILL.md steps 1, 2, and 5) which interact with the host system's terminal.
Sanitization: There is no evidence of sanitization, validation, or escaping of the <file> argument before it is interpolated into the shell command strings.
[COMMAND_EXECUTION]: The skill generates and executes shell commands using variables provided by the user. Specifically, npx playwright test <file> in SKILL.md relies on the <file> variable being a legitimate path. If the agent receives a manipulated string (e.g., test.ts; rm -rf /), it may execute unintended commands on the system.

fix