elastic-beanstalk

Pass

Audited by Gen Agent Trust Hub on Oct 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill analyzes codebase files to determine deployment configuration, which creates a surface for indirect injection from untrusted project files.
  • Ingestion points: The skill reads metadata and configuration files including package.json, requirements.txt, Dockerfile, and application source code to map platforms and determine IAM permissions (SKILL.md, references/platforms.md).
  • Boundary markers: No explicit delimiters or instructions are provided to the agent to treat codebase content as untrusted data during analysis.
  • Capability inventory: The skill utilizes capabilities for AWS CLI command execution, build tool commands (dotnet, mvn, gradle), and cloud environment provisioning.
  • Sanitization: The skill does not define specific sanitization or validation logic for the analyzed codebase inputs before they influence deployment actions.
  • [COMMAND_EXECUTION]: The skill relies on the execution of several CLI tools for environment setup and application building.
  • Evidence: The instructions involve the use of aws elasticbeanstalk commands, dotnet publish, mvn package, and gradle build to prepare and deploy application artifacts (SKILL.md, references/platforms.md).
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 10, 2026, 09:50 AM