The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the agent-team role create command to generate and update local file structures. It passes multiple user-defined strings, including descriptions and goals, as flags to this utility.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user prompts to define role behavior and scope, which is then interpolated into system prompts and command-line arguments. This establishes a surface for malicious instructions to influence the generated role's runtime behavior. * Ingestion points: User-supplied values for role name, description, goals, and scope in SKILL.md. * Boundary markers: The instructions do not define explicit delimiters or 'ignore' instructions for the generated fields. * Capability inventory: The skill can execute shell commands via the agent-team CLI and write to the file system at user-specified target directories. * Sanitization: There are no explicit sanitization or validation routines defined for the user input beyond checking for kebab-case format on the role name.

role-creator