twingate-connectors

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPERSISTENCEPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill documents numerous shell commands and CLI operations for deploying connectors, including Docker operations (docker run, docker-compose), systemd service management (systemctl), and cloud-specific CLI tools (az container, gcloud, aws ecs). These are standard operational requirements for the skill's purpose.
  • [EXTERNAL_DOWNLOADS]: The skill references container images and scripts hosted on binaries.twingate.com, docker.io, ghcr.io, and GitHub. All external repositories (e.g., Twingate-Solutions, Twingate-Labs, Twingate-Community) are officially associated with the vendor and author context.
  • [REMOTE_CODE_EXECUTION]: Installation procedures for several platforms involve fetching and executing remote scripts (e.g., curl | bash). These scripts are sourced from the vendor's verified domains and repositories and are intended for automated product deployment.
  • [PERSISTENCE]: The skill includes instructions for creating systemd services and automated update mechanisms via cron jobs (e.g., a weekly update script). These are legitimate administrative tasks for ensuring service continuity and security compliance.
  • [PRIVILEGE_ESCALATION]: Many recommended commands require sudo or root privileges to configure networking, manage system services, and run containers. The documentation also correctly identifies that certain deployments require privileged: true or host networking to function as a network proxy, providing explicit warnings to the user.
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and summarize a large knowledge base of markdown files. While this creates a potential surface for indirect injection if local files were compromised, the skill provides legitimate technical support based on the provided reference material.
  • Ingestion points: Files within the references/ directory.
  • Boundary markers: Not explicitly defined in the skill logic.
  • Capability inventory: Execution of shell commands and deployment of infrastructure.
  • Sanitization: None specified for the local reference data.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 09:19 PM
Security Audit — agent-trust-hub — twingate-connectors