The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the uip (UiPath) command-line interface to interact with the Maestro registry, manage solutions, and execute project debugging. It also runs python3 for XML well-formedness checks and jq for parsing JSON output from the CLI.
[EXTERNAL_DOWNLOADS]: The instructions guide the installation of the @uipath/cli package from the npm registry, which is a trusted vendor-provided dependency required for the skill's operation.
[DATA_EXFILTRATION]: Project solution data is automatically uploaded to the UiPath cloud infrastructure using the uip solution upload command. Rule 10 specifies that for projects initialized in the current session, this upload occurs without requiring explicit user confirmation to facilitate the build process.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, creating an attack surface for indirect prompt injection. Ingestion points: Parsing and transcription of process images (SKILL.md) and the ingestion of existing .bpmn XML files for editing (references/structural-bpmn.md). Boundary markers: The instructions do not define strict boundary markers or ignore instructions delimiters for this external content. Capability inventory: The skill possesses capabilities including local file writing and execution of shell commands via the Bash tool, as well as network upload via the uip CLI. Sanitization: Rule 9 in the public safety reference provides guidelines for sanitizing outgoing project data to remove sensitive identifiers, though it does not explicitly address the sanitization of incoming data for injection attempts.

uipath-maestro-bpmn