AI Provider Policies

Effective Date: September 16, 2026

Last Updated: September 9, 2026

This document outlines the core policies, terms, and principles for the responsible integration, usage, and development of AI-powered features across Minovative Mind products, including the Minovative Mind CLI (minovative-mind-cli) and VS Code extension. Minovative Mind integrates models strictly and exclusively from Google (Google Cloud Vertex AI and Google AI Studio). All legacy third-party AI providers (including Anthropic PBC / Claude models) have been completely decommissioned and removed from our toolchains and proxy routing. All contributors, maintainers, and users must comply with the referenced policies and best practices to ensure safety, legality, and ethical stewardship of AI technology.

Table of Contents

    1. Responsible AI Principles
    1. Acceptable and Prohibited Uses
    1. Provider-Specific Terms & Privacy
    1. Security, Data Retention, and Abuse Monitoring
    1. Product Integration Guidelines
    1. Reporting Issues and Security Vulnerabilities

1. Responsible AI Principles

Minovative Mind is committed to the ethical development and deployment of AI. We align our integration strategies with the principles established by Google, emphasizing:

  • Safety and accountability
  • Human-centered design
  • Transparency and explainability
  • Privacy and data protection
  • Fairness and prevention of bias

Reference Principles:

2. Acceptable and Prohibited Uses

All project artifacts, integrations, and user interactions with AI models must adhere to the prohibited use policies of our provider. In summary, prohibited uses include (but are not limited to):

  • Unlawful or exploitative activities.
  • Generating harmful, misleading, or deceptive content.
  • Bullying, harassment, or creation of malicious code.
  • Scams, deepfake abuse, and non-consensual data use.
  • Content that violates privacy, intellectual property, or encourages self-harm.

Provider Policies:

3. Provider-Specific Terms & Privacy

By using the AI features in Minovative Mind, you agree to the terms of the provider.

3.1 Google (Vertex AI & Google AI Studio)

Minovative Mind services interface strictly with Google's generative AI infrastructure across managed proxy routing (Google Cloud Vertex AI) and direct client-side Bring Your Own Key (BYOK) configurations (Google AI Studio).

Supported models include:

  • Gemini 3.8 Flash (gemini-3.8-flash) — High-speed flagship reasoning and primary tool-calling model.
  • Gemini 3.7 Flash (gemini-3.7-flash) — Dynamic thinking and multi-agent synthesis model.
  • Gemini 3.6 Flash (gemini-3.6-flash) — Low-latency code transformation and lightweight verification.
  • Gemini 3.1 Pro (gemini-3.1-pro) — Deep multi-modal reasoning and complex architectural planning.
  • Gemini 3.5 Flash-Lite (gemini-3.5-flash-lite) — High-throughput internal auxiliary tasks and classification.

Applicable Provider Terms:

3.2 Decommissioned Providers

All historical third-party providers, including Anthropic (Claude models), have been decommissioned and eliminated from the Minovative Mind ecosystem. No prompts, source code, credentials, or telemetry are transmitted to or processed by Anthropic PBC.

4. Security, Data Retention, and Abuse Monitoring

Minovative Mind utilizes secure enterprise-grade APIs (Google Cloud Vertex AI for managed routing and Google AI Studio for client-side BYOK mode). Notably:

  • Model Training: Prompts, source code, and context submitted to Google AI models via Minovative Mind products (both in BYOK API Key mode and Managed Credit Mode) are not used to train Google foundational models without explicit user opt-in, per Google Cloud enterprise API terms and paid API developer terms.
  • Data Caching: Input and output data may be cached ephemerally by providers for up to 24 hours to improve diagnostics, rate-limiting stability, and inference performance.
  • Managed Mode (Proxy): Workspace context and prompts pass through our secure Google Cloud proxy in transit to Google Cloud Vertex AI. Code context is processed ephemerally and is not persisted on our database servers.
  • BYOK Key Isolation: In BYOK mode, personal Google AI Studio keys reside strictly on the user's local machine within OS-native secure credential stores (or AES-256-GCM encrypted fallback) and communicate directly with Google AI Studio endpoints. Keys and prompts never touch intermediate proxy servers or databases.
  • Session History & Workspace Logs: Conversational transcripts and turn histories are saved strictly client-side within local append-only files (.minovativemind/chat_sessions.jsonl), never retained on Minovative Mind cloud database servers. Context Agent investigation telemetry and session status metadata are recorded in Firestore for auditability, usage tracking, and debugging.

5. Product Integration Guidelines

When contributing to AI integrations in this project, developers must:

  • Secure Credential Management: Never commit API keys or service account credentials to public repositories. Use OS-native keychains (macOS Keychain, Linux libsecret, Windows DPAPI) or AES-256-GCM headless fallback for CLI keys, VS Code's SecretStorage for extension keys, and Google Cloud Secret Manager for backend keys.
  • User Notification: Clearly notify users when AI models generate content, execute commands, run sandboxed probes, or perform autonomous actions.
  • Attribution: Ensure correct attribution and model identification in the UI using Google's canonical naming convention (e.g., "Powered by Google Cloud Vertex AI / Gemini 3.8 Flash").
  • Safety Filtering: Implement and respect the safety settings provided by the Google AI APIs to filter out harmful content.
  • Model String Hygiene: Avoid hardcoding deprecated model IDs. Strictly use active Gemini 3.x series models (gemini-3.8-flash, gemini-3.7-flash, gemini-3.6-flash, gemini-3.1-pro, gemini-3.5-flash-lite).

6. Reporting Issues and Security Vulnerabilities

All security concerns or potential AI policy violations should be reported through the channels below:

Disclaimer

This document is a summary of the official policies of our AI provider and must be used in conjunction with the latest policy links provided above. Official provider terms always take precedence. Model availability, pricing, and terms are subject to change; always verify against the provider's current documentation before deploying to production.