uipath-genome
Warn
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill implements a mechanism to clone 'framework migration packs' from user-provided git URLs or folder paths.
- If a git URL is supplied, the skill clones it into a scratch folder before processing.
- [REMOTE_CODE_EXECUTION]: Scripts contained within external framework migration packs are executed during the genome extraction process.
- The skill instructs the agent to 'Run the row's script before reading any export file', which allows for the execution of arbitrary code provided in the migration pack.
- [COMMAND_EXECUTION]: The skill frequently executes system commands and local scripts to perform its automation tasks.
- It utilizes the UiPath CLI (
uip) for platform operations, resource management, and UI automation tasks. - It runs a local Python script
scripts/check-genome.pyto validate the generated genome files. - It uses
sedto filter and process XAML workflow files. - [INDIRECT_PROMPT_INJECTION]: The skill has a high surface area for indirect prompt injection as it parses complex automation projects to extract logic.
- Ingestion points: Reads numerous file types including XAML, BPMN, JSON, and custom framework exports (e.g., in
uipath-source-guide.md). - Boundary markers: No explicit instruction delimiters or 'ignore' warnings for the content of processed project files are mentioned.
- Capability inventory: Broad capabilities including CLI subprocess execution, file writing, and network operations via git cloning.
- Sanitization: The skill provides detailed guidance for scrubbing secrets and redacting password-looking fields from derived data (e.g., in
source-migration-guide.md). - [DYNAMIC_EXECUTION]: The skill uses runtime-executed scripts to generate inventories and catalogs needed for migration.
- It relies on 'inventory scripts' from framework migration packs to render the source projects into a format the agent can process.
Audit Metadata