colab-operator

Pass

Audited by Gen Agent Trust Hub on Jun 19, 2026

Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill makes extensive use of local shell commands to manage the environment:
  • Uses gcloud auth application-default login to configure Google Cloud credentials.
  • Uses the colab CLI tool for provisioning and lifecycle management of virtual machines.
  • [REMOTE_CODE_EXECUTION]: The skill's primary purpose is to execute arbitrary code on remote infrastructure:
  • Uses colab exec and colab run to transmit and execute local Python scripts on a remote Colab kernel.
  • Uses colab console to execute shell commands on the remote VM.
  • [CREDENTIALS_UNSAFE]: The skill instructs the agent to configure authentication using high-privilege scopes:
  • Specifically requests the https://www.googleapis.com/auth/cloud-platform scope, which grants broad access to all Google Cloud resources in the user's project.
  • Manages sensitive authentication tokens stored in ~/.config/colab-cli/token.json and session state in ~/.config/colab-cli/sessions.json.
  • [EXTERNAL_DOWNLOADS]: The skill relies on an external CLI tool and instructs users to install it via standard package managers:
  • Recommends installation of the google-colab-cli package using pip or uv.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 19, 2026, 11:18 AM
Security Audit — agent-trust-hub — colab-operator