commerce-app-storage
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install official Adobe libraries and CLI plugins.
- Evidence:
- Installation of
@adobe/aio-lib-dbfrom the standard registry. - Installation of the
@adobe/aio-cli-plugin-app-storageplugin from the official Adobe repository (github.com/adobe/aio-cli-plugin-app-storage). - These are trusted resources belonging to the skill's authoring organization (Adobe).
- [COMMAND_EXECUTION]: The skill uses standard development commands and Adobe-specific CLI tools to initialize and provision the project environment.
- Evidence:
- Execution of
npx @adobe/aio-commerce-lib-app initto scaffold the project structure. - Use of
aio app db provisionto set up workspace storage. - These commands are routine for Adobe Commerce app development and operate within the project's development context.
- [INDIRECT_PROMPT_INJECTION]: The skill describes a workflow where the agent ingests user intent to generate functional code, which is a common surface for indirect instructions.
- Ingestion points: The agent gathers 'Action type', 'Collection name', and 'Operations' from the user in Step 3 of
SKILL.md. - Boundary markers: The skill provides strict structural requirements for registration (e.g.,
include-ims-credentials: true) and code implementation (init -> connect -> close) which serve as behavioral constraints for the generated output. - Capability inventory: The generated code performs database operations (read/write) using authenticated sessions and handles runtime actions through the
aio-lib-dbclient. - Sanitization: The instructions include specific guidance on handling input safely, such as properly converting string IDs to
ObjectIdusing thebsonlibrary and using projections to limit data exposure. - [SAFE]: The skill follows security best practices for secret management and resource lifecycle.
- Evidence:
- Instructions mandate using
finallyblocks to ensure database connections are closed, preventing resource exhaustion. - Authentication relies on platform-injected IMS credentials (
AIO_COMMERCE_AUTH_IMS_*) rather than hardcoded secrets or unsafe credential storage.
Audit Metadata