litium-cloud-migration

Pass

Audited by Gen Agent Trust Hub on Sep 30, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONCREDENTIALS_UNSAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill analyzes external content from the customer's repository, including configuration files, project files, and a dedicated state file (MIGRATION.md). This ingestion of untrusted data, combined with the ability to generate manifests and execute cloud management commands, creates a surface for indirect prompt injection.
  • Ingestion points: Reads and interprets MIGRATION.md to track state across sessions; scans source code and configuration files via grep and find to inventory the site architecture.
  • Boundary markers: The skill relies on natural language instructions for the agent to distinguish between its own logic and the data found in the repository, but lacks formal delimiters for ingested content.
  • Capability inventory: The skill produces deployment manifests, pipeline configuration files, and executes commands through a cloud CLI tool for environment management, artifact creation, and deployment.
  • Sanitization: No specific sanitization or filtering logic is described for the data extracted from the repository before it is processed by the agent.
  • [EXTERNAL_DOWNLOADS]: The skill guides the user to download and install official tools and packages from verified vendor sources.
  • Fetches a cloud CLI tool as a global .NET tool during the build and deployment phases.
  • Downloads deployment and migration references from the vendor's official documentation domains.
  • References specific build-time packages for generating cloud-native deployment manifests.
  • [COMMAND_EXECUTION]: The skill provides numerous shell commands for repository analysis, building artifacts, and managing cloud infrastructure.
  • Uses command-line utilities to perform automated inventory of dependencies, platform-specific code, and file system paths.
  • Executes build commands to package the application for specific operating system targets.
  • Orchestrates the migration process using a command-line interface for environment setup and application deployment.
  • [CREDENTIALS_UNSAFE]: The skill provides templates for CI/CD pipelines and manual configuration that handle sensitive credentials, though it emphasizes secure practices.
  • Requires service principal identities and certificates for automated deployments in continuous integration environments.
  • Manages secrets for external integrations (such as email and payment services) using an indirect reference mechanism in configuration manifests.
  • Explicitly instructs users to avoid committing secrets to the repository and provides guidance on using secure secret stores.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 30, 2026, 11:39 AM