The Agent Skills Directory

[SAFE]: The skill implements strict operational boundaries via 'Critical Rules' which explicitly forbid the use of external network tools (curl, wget) and direct REST API calls, ensuring all operations are performed through the authenticated uip ixp CLI tool.
[SAFE]: To mitigate risks from indirect prompt injection (Category 8) while processing untrusted documents, the skill defines a rigorous review process. The agent is instructed to act as a 'reviewer' rather than an 'extractor,' performing field-by-field verification against the document content before issuing confirmation or correction commands.
[SAFE]: The skill handles file management in a safe manner, using a localized temporary directory (/tmp/ixp/) and specific instructions to avoid platform-specific path resolution issues (e.g., between Windows and bash environments).
[SAFE]: Destructive operations such as project, group, or document deletion are correctly documented with the requirement of mandatory confirmation flags (-y, --yes), preventing accidental data loss or unauthorized modifications.
[SAFE]: All external documentation links point to official docs.uipath.com domains, and the required CLI tool is a recognized resource from the verified vendor 'uipath'.

uipath-ixp