sealos-database

Pass

Audited by Gen Agent Trust Hub on Aug 7, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes the official sealos-cli for database provisioning and lifecycle management. It also runs a local utility script analyze-project-database.mjs to detect project requirements.
  • [REMOTE_CODE_EXECUTION]: If sealos-cli is not found, the skill invokes npx -y sealos-cli@latest, which downloads and runs the vendor's official CLI tool from the public npm registry.
  • [EXTERNAL_DOWNLOADS]: The skill downloads the sealos-cli package via npx and communicates with the Sealos Cloud API endpoints to manage infrastructure.
  • [DATA_EXFILTRATION]: The skill is designed to handle sensitive connection details but includes explicit safety constraints that prohibit the agent from displaying secrets in responses or committing them to files.
  • [PROMPT_INJECTION]: The skill processes project files through analyze-project-database.mjs, creating an indirect prompt injection surface. Ingestion points: analyze-project-database.mjs reads text files in the project root. Boundary markers: The script returns structured JSON output. Capability inventory: The skill can manage cloud databases and modify local environment files. Sanitization: The script uses regular expressions for pattern matching but does not sanitize content before inclusion in the JSON report.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 7, 2026, 07:29 AM
Security Audit — agent-trust-hub — sealos-database