The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Risk. The skill ingests untrusted data from the UiPath Audit Service (audit logs) which contains metadata that could be controlled by an attacker.
Ingestion points: The uip admin audit <scope> events command, described in references/audit-commands.md, retrieves event logs containing potentially malicious strings in fields like eventDetails or eventSummary.
Boundary markers: The skill instructions emphasize using --output json for programmatic parsing, providing structural separation between data and commands.
Capability inventory: The agent has access to the Bash tool and the uip admin CLI, which allows for significant mutations such as user management, role assignments, and tenant lifecycle operations.
Sanitization: No explicit instructions are provided to sanitize or escape the content of the audit logs before they are displayed or processed further.
[SAFE]: The skill incorporates extensive security best practices and guardrails.
Critical Rule 5 mandates the resolution and verification of principal IDs (Users, Groups, Robot Accounts) before any administrative mutation is performed.
Critical Rule 31 enforces a mandatory impact statement and explicit user confirmation before enabling IP restriction enforcement to prevent accidental lockout.
The skill consistently follows a "Discover before creating" pattern to avoid duplicates and ensure operational integrity.
All operations prioritize structured output (--output json) for reliable programmatic handling.

uipath-admin