The Agent Skills Directory

[COMMAND_EXECUTION]: The skill defines a tool capability Bash(npm test:*) and instructs the agent to accept user-provided arguments for specific files or modules. There is no evidence of input validation or shell-escaping in the instructions, which creates a potential command injection surface. An attacker could provide a string containing shell metacharacters (such as ;, &&, or |) to execute unauthorized commands on the system.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it reads and analyzes local project files using Read, Grep, and Glob to generate testing recommendations. Maliciously crafted content within the analyzed source code, comments, or documentation could influence the AI's analysis or steer it toward recommending unsafe actions.
Ingestion points: Project source files and modules accessed via Read, Grep, and Glob.
Boundary markers: Absent; the skill does not use delimiters or instructions to ignore embedded commands in analyzed content.
Capability inventory: The skill possesses the ability to execute shell commands (Bash) and access the file system.
Sanitization: No sanitization or validation logic is defined for the content ingested from the codebase.

coverage