The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to run npx playwright test using a file path provided in $ARGUMENTS. The instructions do not specify that the agent should quote or sanitize the <file> variable, which creates a command injection risk if the user-supplied path contains shell metacharacters like semicolons or pipes.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input from $ARGUMENTS and interpolates it into sensitive instructions and shell commands without protective measures.\n
Ingestion points: The $ARGUMENTS object in SKILL.md containing test file paths, names, and descriptions.\n
Boundary markers: Absent. There are no delimiters or instructions provided to the agent to treat the input as non-executable data.\n
Capability inventory: The skill possesses capabilities for shell command execution via npx and local filesystem access for reading flaky-taxonomy.md.\n
Sanitization: Absent. No escaping, validation, or filtering of the user-provided arguments is defined in the skill logic.

fix