Privacy Policy for Minovative Mind CLI

Effective Date: September 16, 2026

Last Updated: September 9, 2026

Minovative Mind ("we", "us", or "our") respects your privacy and is committed to protecting the personal and project information you share with us while using the Minovative Mind CLI (the "Service"). This Privacy Policy explains what information we collect, how we use it, how it is shared, and your choices regarding that information.

1. Information We Collect

When you use the Minovative Mind CLI, we may collect the following types of information:

A. Account and Authentication Information

To use the Service, you must authenticate via GitHub. When you run minovative-mind-cli login, we collect basic profile information provided by the GitHub OAuth API, which may include:

  • Your GitHub username and user ID
  • Your primary email address associated with your GitHub account
  • An OAuth access token to authenticate your requests to our backend services

OS-Native Secure Credential Storage & AES-256-GCM Fallback: Your authentication credentials, Firebase User IDs (UIDs), and refresh tokens are stored locally on your machine using OS-native secure credential stores (macOS Keychain, Linux libsecret, Windows DPAPI / Credential Manager) where available, or in an AES-256-GCM encrypted fallback file (.credentials.enc) with owner-only (0600) permissions in headless or non-GUI environments. Personal API keys used for Bring Your Own Key (BYOK) mode (e.g., Google AI Studio keys configured via /config-key) are stored strictly within your local secure credential store and are never transmitted to, logged by, or retained on Minovative Mind database servers.

B. Workspace Metadata, Code Context & Subscription Verification Data

The Minovative Mind CLI is an intelligent autonomous coding agent that reads files and context from your local workspace to assist with software engineering tasks.

  • Workspace Metadata Transmission: When you initialize, register, or use a workspace, the CLI transmits absolute local directory paths (e.g., /Users/username/projects/my-app), workspace names, and your user UID to Google Firestore (hosted via Firebase). This metadata is stored on our database servers to map and manage your workspaces, track active projects, and link usage to your account.
  • BYOK Subscription Verification: To use your own Google AI Studio API key in BYOK mode, an active $3.99/month BYOK Subscription is required to cover software infrastructure, multi-workspace routing, and account maintenance. When executing in BYOK mode or configuring keys via /config-key, the CLI transmits your Firebase authentication ID token over HTTPS to our verifySubscription Cloud Function to confirm active subscription status in Firestore (/customers/{uid}/subscriptions). No prompt text, code context, or personal API keys are transmitted during subscription verification.
  • Transmitted Code Context Data: When you interact with the agent (e.g., via minovative-mind-cli chat or autonomous sub-agent orchestration), the CLI bundles your prompts, relevant file contents, error messages, diffs, and terminal outputs, and sends them over encrypted TLS/HTTPS connections to our custom proxy endpoint, which routes securely to our AI inference providers (such as Google Cloud Vertex AI or Google AI Studio) to generate completions and tool actions.
  • Local Client-Side Append-Only JSONL Chat Persistence: When interacting with the agent, conversational session turns, messages, and tool invocations are persisted client-side on your local filesystem within your workspace's .minovativemind/chat_sessions.jsonl file. This engine uses an append-only JSON Lines format with local in-memory byte offset indexing for fast random access, serialized FIFO write queuing, tombstone deletion, and automated background compaction (>40% dead records). The .minovativemind/ directory is automatically appended to your project's .gitignore to prevent conversational transcripts from ever entering version control.
  • Zero Cloud Retention of Code Context & Transcripts: While workspace directory paths and workspace names are stored in Firestore for project mapping, we do not store the contents of your source code, proprietary repository files, prompts, or conversational chat transcripts on Minovative Mind cloud database servers. Code context data is processed ephemerally by third-party AI models during the active request lifecycle and is not retained by Minovative Mind once inference completes.

C. Usage Data and Telemetry

To improve the Service, we collect anonymized usage statistics and telemetry data, including:

  • CLI commands executed (e.g., chat, login, logout, eval, models, sub-agents, profile)
  • AI model usage statistics (e.g., tokens sent/received, response latency, model selected)
  • System information (e.g., Operating System, Node.js version, CLI version)
  • Crash reports and error logs (to help us debug and fix issues)

D. Adaptive User Profile & Local Persona Memory

To deliver a deeply personalized developer experience, the CLI features an adaptive memory engine that observes your interaction patterns, communication style, cognitive problem-solving habits, and technical conventions:

  • Data Categories Observed:
    • Communication Traits: Tone, verbosity, and thought formulation/prompting style.
    • Cognitive & Decision-Making Spectrums: Architectural orientation (top-down vs. bottom-up), decision autonomy (direct recommendations vs. option exploration), risk and velocity preferences (pragmatic speed vs. defensive rigor), delegation depth, debugging style, and explanation formats.
    • Conversational Persona & Teammate Dynamics: Pairing relationship models (collaborative peer vs. direct operator), banter and humor affinity, speech formality, apology and error reaction, stress and urgency cadence, and conversational quirks/catchphrases.
    • Technical Conventions: Language proficiencies, architectural patterns, framework paradigms, and coding conventions.
    • AI Observations & Side-Notes: Incremental observations and contextual notes recorded by the agent to tailor future responses.
  • Local Storage Only: All adaptive persona traits, cognitive spectrums, and agent side-notes are stored strictly on your local machine at ~/.minovativemind/user_profile.json with owner-only (0600) permissions.
  • No Cloud Database Retention: User profile side-notes, persona traits, and cognitive preferences are never transmitted to, synced with, or retained on Minovative Mind backend database servers.
  • Full Transparency and User Control: You maintain complete ownership and control over your profile. You can inspect the exact observations recorded by the agent, delete individual side-notes, or permanently reset your entire profile memory at any time using the /profile slash command.

2. How We Use Your Information

We use the information we collect for the following purposes:

  • To Provide the Service: To authenticate you, manage active workspaces, map project paths, verify subscription status, process your coding requests, and deliver AI-generated code and command suggestions to your terminal.
  • To Maintain and Improve the Service: To debug issues, analyze aggregate usage patterns, optimize inference performance, and develop new features for the CLI.
  • To Communicate with You: To send important technical notices, security alerts, or updates regarding the Terms of Use and Privacy Policy.

3. Data Sharing and Third-Party Subprocessors

We do not sell your personal information or your source code to third parties. However, we share data with trusted third-party service providers necessary to operate the Service:

  • Google Cloud Run & Vertex AI (Google LLC): We host our API/proxy endpoints on Google Cloud Run and use Google Cloud Vertex AI to process your prompts and analyze local code context using the Gemini foundation model family—including gemini-3.8-flash, gemini-3.7-flash, gemini-3.6-flash, gemini-3.1-pro, and gemini-3.5-flash-lite. As per Google Cloud's enterprise privacy and data governance commitments, data sent to Vertex AI via the API is not used by Google to train foundational models. The Minovative Mind CLI operates exclusively on Google Gemini foundation models.
  • Google AI Studio (Google LLC): When operating in Bring Your Own Key (BYOK) mode with a user-supplied API key, inference requests are sent directly or routed securely to Google AI Studio under your own Google Developer credentials, subject to Google's Generative AI Terms of Service and Developer API policies.
  • Google Firebase & Cloud Firestore (Google LLC): We store your user UID, workspace names, and absolute local directory paths in Google Firestore to maintain workspace registry state, account subscription status, and credit quotas.
  • GitHub (GitHub, Inc. / Microsoft Corp.): Used exclusively for developer identity verification and OAuth authentication.

4. Data Security

We implement rigorous technical and organizational measures to safeguard your credentials, workspace metadata, and local development environment:

  • Encryption in Transit: All communications between the CLI, Minovative Mind backend proxy services, Firebase Firestore, and third-party AI inference providers (Google Cloud Vertex AI, Google AI Studio) are strictly encrypted in transit using industry-standard TLS 1.3 / HTTPS protocols.
  • OS-Native Secure Credential Storage: Authentication credentials, session tokens, and Firebase UIDs are stored in operating system-level secure credential vaults wherever supported—including macOS Keychain, Linux Secret Service (libsecret), and Windows Data Protection API (DPAPI) / Credential Manager.
  • AES-256-GCM Encrypted Fallback: In headless server environments, CI/CD pipelines, or systems where native OS keychains are unavailable, credentials and sensitive configurations are encrypted at rest using AES-256-GCM authenticated encryption and stored in .credentials.enc with strict owner-only (0600) file permissions.
  • Local BYOK Key Isolation: Personal API keys supplied for Bring Your Own Key (BYOK) mode (such as Google AI Studio keys configured via /config-key) reside exclusively on your local machine within the secure credential subsystem. Personal keys are never transmitted to, logged by, or stored within Minovative Mind backend databases.
  • Local Adaptive Profile Protection: Adaptive user profile data, communication traits, cognitive spectrums, and agent observations are isolated at ~/.minovativemind/user_profile.json with owner-only (0600) permissions and are never uploaded to or synchronized with cloud databases.
  • Local Project Storage & Chat Session Isolation: Chat history transcripts (.minovativemind/chat_sessions.jsonl) and workspace dependency caches reside exclusively within your local workspace's .minovativemind directory, with automated .gitignore rules preventing unintentional commits to source control.
  • Isolated Execution Sandboxes: Autonomous diagnostic probes, dynamic test-driven generation, memory audits, and formal mathematical verifications (run_debug_script, run_fuzz_probe, check_heap_delta, check_behavioral_drift, and verify_formal_proof) execute in isolated child processes and sandboxed operating system temporary directories (os.tmpdir()). These sandboxes are bounded by strict process execution timeouts (2 minutes for runtime scripts and fuzz probes; 3 minutes for Lean 4 formal proof verification) to prevent infinite loops, resource exhaustion, and unauthorized workspace file mutations. Formal verification incorporates anti-cheating guardrails that block proof-bypassing escape hatches (such as sorry, admit, unsafe, #check_failure, or unproven axioms) and strictly prohibits modifying verification toolchain configurations (lakefile.lean, lakefile.toml, lean-toolchain).
  • Local User Account Security: While credentials, keys, and profile memories are protected through native OS enclaves, encrypted storage, and process sandboxes, you remain responsible for maintaining the physical and logical security of your local operating system user account and development machine.

5. Your Responsibilities Regarding Sensitive Data

Because the CLI reads your local workspace and transmits it to an LLM for processing, you are strictly responsible for ensuring you do not pass sensitive information to the AI. You must ensure that:

  • Hardcoded API keys, passwords, and database credentials are excluded from the context.
  • Personally Identifiable Information (PII) or Protected Health Information (PHI) is not present in the files the CLI inspects.
  • You utilize .gitignore or CLI-specific ignore files (such as .minovativeignore) to prevent the CLI from reading .env files, private keys, or other secret-containing directories.

6. Data Retention

  • Account and Database Metadata: We retain your authentication metadata and workspace metadata (such as directory paths and workspace names stored in Firestore) for as long as your account is active or as needed to provide the Service.
  • Ephemeral Code Context & Zero Cloud Transcript Retention: As stated above, we do not store your code context, repository source files, prompt bundles, or conversational chat transcripts on our cloud database servers. Prompts are processed ephemerally by third-party AI models during active inference and immediately discarded. Chat session transcripts are retained strictly client-side on your local device within .minovativemind/chat_sessions.jsonl.
  • Telemetry Data: Anonymized usage statistics and crash reports are retained for a period necessary to analyze and improve the Service.

7. Your Choices and Rights

Depending on your location, you may have certain rights regarding your personal information, including the right to access, correct, or delete your data.

  • Account Deletion / Logout: You can revoke access at any time by running minovative-mind-cli logout and revoking the OAuth application from your GitHub account settings.
  • Purging Local Chat Sessions and Project Caches: You have the complete right and autonomous ability to purge local chat session histories and workspace caches at any time. You can delete the .minovativemind/chat_sessions.jsonl file or remove the entire .minovativemind/ directory from your workspace root. Deleting these local files immediately erases all stored conversation turns, indexing tables, and cached symbols without impacting your cloud account.
  • Managing & Deleting Adaptive Persona Data: You can inspect, selectively delete individual side-notes, or permanently reset your entire adaptive persona memory at any time by typing /profile in an interactive chat session.
  • Opting Out of Telemetry: You can opt out of anonymous telemetry collection by setting an environment variable (e.g., MINOVATIVE_TELEMETRY=0) or updating your CLI configuration. Please refer to the documentation for specific instructions.

8. Cross-Border Data Transfers

Minovative Mind operates globally, and our backend infrastructure (including Google Cloud Run and Google Firestore) is primarily hosted in the United States. By using the Service, you acknowledge and agree that your workspace metadata, telemetry, and ephemerally processed code context may be transferred to, stored, and processed in the United States or other countries where our third-party subprocessors operate. We take appropriate safeguards to ensure that your data remains protected in accordance with this Privacy Policy, regardless of where it is processed.

9. Changes to this Privacy Policy

We may update this Privacy Policy from time to time to reflect changes in our practices, architectural enhancements, or legal obligations. We will notify you of any material changes by updating the "Effective Date" and, when feasible, providing a notification within the CLI. Your continued use of the Service after the effective date constitutes your acceptance of the updated Privacy Policy.

10. Contact Us

Security Reporting Form: https://forms.gle/QexZY2resdXpahUK6