catalyst-smartbrowz
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides a surface for indirect prompt injection by processing untrusted external data.
- Ingestion points: The skill ingests external content via methods such as smartbrowz.convertToPdf(), smart_browz.take_screenshot(), smartbrowz.generateFromTemplate(), and Dataverse modules like getEnrichedLead and findTechStack in references/smartbrowz-basics.md.
- Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from following instructions potentially embedded in the processed data.
- Capability inventory: The skill allows for network operations to access arbitrary URLs and file system writes for generating PDF and image files.
- Sanitization: The documentation does not specify any sanitization or validation of the input HTML, URLs, or template data before processing.
- [COMMAND_EXECUTION]: The skill requires the Catalyst CLI for project initialization and deployment.
- Evidence: Commands like catalyst init, catalyst functions:create, and catalyst deploy are documented in references/smartbrowz-basics.md.
- [DATA_EXFILTRATION]: The skill includes components for web scraping and organizational data extraction from external sources.
- Evidence: The Dataverse component in references/smartbrowz-basics.md provides modules for lead enrichment and tech stack discovery from external domains.
- [DYNAMIC_EXECUTION]: The skill performs runtime rendering of user-provided HTML and templates.
- Evidence: The SDK methods convertToPdf and generateFromTemplate in references/smartbrowz-basics.md dynamically execute rendering logic on user-provided strings.
Audit Metadata