The Agent Skills Directory

[SAFE]: The skill provides instructions for using the nooa command-line tool and library for model configuration. It explicitly mandates security best practices, including:
Storing sensitive keys in environment variables or dedicated secrets files rather than configuration files or shell arguments.
Requiring explicit user consent and allowance before performing operations that incur financial costs (paid inference calls).
Validating configurations through an offline 'plan' stage before persistence.
Sanitizing handoff reports to avoid leaking reasoning text, encrypted state, or raw error bodies.
[COMMAND_EXECUTION]: The skill describes the use of the nooa CLI tool (e.g., uv run nooa connect). These commands are standard for the tool's intended purpose of model registration and management and are presented with appropriate warnings regarding user authorization.
[INDIRECT_PROMPT_INJECTION]: The skill processes external data via configuration files and command-line arguments. It includes appropriate defensive measures, such as explicit boundary markers (user approval steps) and sanitization guidelines for output reports, to mitigate potential injection risks from untrusted configuration inputs.

nooa-model-configuration