skills/q00/ouroboros/config/Gen Agent Trust Hub

config

Warn

Audited by Gen Agent Trust Hub on Jul 31, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The instructions direct the agent to execute ouroboros config --web --host 0.0.0.0 --no-browser when operating in a remote host environment. Binding a service to 0.0.0.0 allows it to accept connections from any IP address reachable by the host, potentially exposing the settings GUI to unauthorized network actors.
  • [EXTERNAL_DOWNLOADS]: The skill advises the agent to suggest running pip install 'ouroboros-ai[tui]' if the command fails due to missing dependencies. This involves the installation of third-party software from a public repository into the user's execution environment.
  • [DATA_EXFILTRATION]: The skill manages access to ~/.ouroboros/config.yaml, a configuration file containing runtime profiles and model selections. While the skill's purpose is configuration, the ability to serve this data over a network-accessible web server (0.0.0.0) creates a path for sensitive environment data to be accessed externally.
  • [PROMPT_INJECTION]: The skill contains a 'RFC #1392 State Breadcrumb Footer' requirement, instructing the agent that its 'final response MUST end with exactly one breadcrumb footer line'. This is a form of instruction overriding that dictates strict constraints on the agent's output behavior and format.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 31, 2026, 04:17 PM
Security Audit — agent-trust-hub — config