The Agent Skills Directory

[DATA_EXFILTRATION]: The skill manages sensitive authentication tokens stored in ~/.uipath/.auth and profile-specific configuration paths. Access to these files is necessary for the CLI to authenticate with UiPath Cloud services, representing a sensitive data surface.
[DYNAMIC_EXECUTION]: The skill documents an 'Activity Generation' workflow that involves generating JavaScript scripts from vendor API specifications and executing them locally using the uip is resources run script command. This dynamic code generation is a primary feature for extending platform connectors.
[INDIRECT_PROMPT_INJECTION]: The skill has a significant exposure to indirect prompt injection through its ingestion of untrusted data from various external vendor APIs and the Data Fabric storage service.
Ingestion points: Untrusted data enters the agent context through responses from Integration Service connectors, record queries from Data Fabric entities, and execution logs from Orchestrator jobs.
Boundary markers: The skill utilizes structured JSON objects and CEQL filter trees for data processing, though it lacks explicit natural language boundary markers for all processed content.
Capability inventory: The skill is capable of executing shell commands, writing files, and invoking additional agent skills.
Sanitization: Instructions prioritize the use of structured data formats and server-side filtering to mitigate risks associated with untrusted external content.
[REMOTE_CODE_EXECUTION]: The skill provides installation commands that download and execute scripts from download.uipath.com. These are official vendor installation paths and are considered safe within the context of the skill author.
[COMMAND_EXECUTION]: The skill uses the Bash tool to manage environment setup and monitor platform resources via the uip CLI.

uipath-platform