The Agent Skills Directory

[COMMAND_EXECUTION]: Employs the uip CLI and npm for task orchestration, solution scaffolding, and flow validation. It also includes shell snippets for environment discovery and sourcing vendor configurations.
[EXTERNAL_DOWNLOADS]: Interacts with official uipath.com domains to search for and configure Action Center applications, which is essential for synchronizing with the vendor's cloud resources.
[CREDENTIALS_UNSAFE]: Accesses the official vendor configuration file located at ~/.uipath/.auth to retrieve authentication tokens required for CLI operations. This is standard behavior for the UiPath development environment.
[INDIRECT_PROMPT_INJECTION]: Manages untrusted data ingestion via human reviewer inputs and sidecar schema files. The skill implements security controls by mandating validation through the uip maestro toolset.
Ingestion points: Human responses in forms and .hitl.json schema definitions.
Boundary markers: Use of structured JSON data with explicitly defined field roles.
Capability inventory: Execution of shell commands via the uip CLI and local file system modifications.
Sanitization: Uses platform-native validation commands to ensure schema and wiring integrity prior to execution.

uipath-human-in-the-loop