package-usage
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions direct the agent to use the
pulumi apiCLI command to retrieve package and stack metadata. This is a standard administrative operation for the Pulumi platform and is executed within the context of the user's existing Pulumi authentication. - [INDIRECT_PROMPT_INJECTION]: The skill ingests and processes JSON data returned from the Pulumi Cloud API (specifically stack names, project names, and version strings).
- Ingestion points: API responses from
pulumi api /api/registry/packagesandpulumi api /api/orgs/{orgName}/packages/usagedescribed inSKILL.md. - Boundary markers: None explicitly defined for the API output processing.
- Capability inventory: The skill is limited to formatting data into markdown tables; it does not possess file-writing, arbitrary code execution, or non-vendor network capabilities that could be exploited via malicious API data.
- Sanitization: No specific sanitization is mentioned, but the risk is mitigated by the skill's restricted scope of action (version comparison and output formatting).
Audit Metadata