devops

Pass

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCREDENTIALS_UNSAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides installation instructions for the platform CLI by piping remote shell scripts and PowerShell scripts to interpreters (curl | bash and irm | iex). These downloads originate from the official project repository of a well-known DevOps service.
  • [EXTERNAL_DOWNLOADS]: Fetches the Coolify CLI binary and installer scripts from the project's official GitHub organization (raw.githubusercontent.com/coollabsio/).
  • [CREDENTIALS_UNSAFE]: The skill manages sensitive configuration and authentication. It identifies the location of platform configuration files (~/.config/coolify/config.json) which contain API tokens and provides examples of adding SSH private keys (e.g., ~/.ssh/id_rsa) to the management platform. These actions are standard for the tool's primary purpose of DevOps management.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection as it ingests and processes unvalidated data from external sources.
  • Ingestion points: Reads application logs, server status reports, and deployment details (found in coolify/SKILL.md).
  • Boundary markers: Lacks explicit delimiters or instructions to ignore commands embedded within ingested log or status data.
  • Capability inventory: Possesses significant infrastructure mutation capabilities, including resource deletion, database lifecycle management, and server removal.
  • Sanitization: Does not specify filtering or escaping mechanisms for external content before processing.
  • [COMMAND_EXECUTION]: Executes the coolify CLI tool to perform all infrastructure management tasks and platform queries.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 13, 2026, 09:20 AM