The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it incorporates untrusted input into parameters for Harness management tools. Malicious instructions embedded in user data could be executed by the agent.
Ingestion points: Untrusted data enters via the search_term parameter in harness_list and various fields in the body parameter for harness_create as seen in SKILL.md.
Boundary markers: No delimiters or warnings are used to differentiate between instructions and user-supplied data.
Capability inventory: The skill employs harness_list, harness_get, harness_create, and harness_delete in SKILL.md to perform administrative actions on users, groups, and service accounts.
Sanitization: The instructions do not define any sanitization or validation for the interpolated data.

manage-users