using-marketplace-templates
Pass
Audited by Gen Agent Trust Hub on Oct 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the ingestion of external data from the Celigo Marketplace, creating a potential attack surface.
- Ingestion points: The agent reads template previews, author-provided readmes, and catalog listings via
celigo templates previewandlist_marketplacecommands (inSKILL.md). - Boundary markers: The instructions lack explicit boundary markers or directives to treat external template content as untrusted data that should not influence agent behavior.
- Capability inventory: The skill grants the agent extensive capabilities via the
celigoCLI, including integration installation, connection mapping, script management, and MCP server configuration. - Sanitization: There is no mention of sanitizing or validating the content of marketplace blueprints before the agent interprets their instructions.
- [COMMAND_EXECUTION]: The skill relies on shell command execution via the
celigoCLI to interact with the platform. While these are vendor-provided tools, they perform sensitive operations such as account searching, integration management, and credential mapping.
Audit Metadata