autumn-integrate

Pass

Audited by Gen Agent Trust Hub on Sep 22, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill instructs the installation of official vendor packages, including autumn-js, autumn-sdk, and @useautumn/sdk, through standard package registries like npm and pip.
  • [COMMAND_EXECUTION]: The skill utilizes the atmn CLI tool for project initialization and configuration management, including commands like atmn init and atmn push to synchronize billing plans.
  • [DATA_EXFILTRATION]: The skill facilitates the transmission of customer metadata, such as IDs, names, and emails, to the vendor's official API at api.useautumn.com. This behavior is documented and inherent to the billing service functionality.
  • [INDIRECT_PROMPT_INJECTION]: The skill includes an attack surface where external data is ingested for processing.
  • Ingestion points: Customer identifiers, names, and email addresses are ingested from the application's authentication provider as seen in references/creating-customers.md.
  • Boundary markers: No explicit boundary markers or delimiters for user-supplied data are mentioned in the integration examples.
  • Capability inventory: The skill uses network operations to communicate with api.useautumn.com and writes local configuration files via the atmn CLI tool as described in references/setup.md.
  • Sanitization: There is no specific instruction for sanitizing or validating the customer-provided strings before they are passed to the SDK methods in the provided code snippets.
  • [SAFE]: No malicious activities, obfuscated code, or unauthorized persistence mechanisms were found. The skill demonstrates safe practices by recommending the use of .env files for managing the AUTUMN_SECRET_KEY.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 22, 2026, 03:47 AM
Security Audit — agent-trust-hub — autumn-integrate