DiSkribe is an AI-powered assistant specialised in the DiSky Skript add-on for Minecraft servers. It can answer questions about DiSky syntax, Discord bot development, event handling, and more — grounding every answer in the official DiSky documentation so you always get accurate, up-to-date code.

DiSky is a Skript add-on that lets you build fully-featured Discord bots directly from Minecraft server scripts. It covers slash commands, embeds, modals, select menus, voice channels, threads, and much more. You can find the official documentation at the DiSky wiki.

Yes. DiSkribe uses Discord OAuth for authentication — no separate password required. Your Discord username and avatar are displayed in the app; your email is stored server-side only for account management purposes and is never shared publicly.

DiSkribe supports multiple AI models and providers. The available models depend on your subscription tier and the administrator configuration. Typically you will find:

• Fast models — lightweight, low-latency, ideal for simple questions and quick code snippets.

• Smart / reasoning models — more powerful, capable of extended thinking for complex debugging tasks. Available on Premium and Platinum tiers.

You can switch models at any time using the model chip in the chat composer.

Yes. While DiSkribe actively queries the DiSky wiki and syntax index before every answer, AI models can still misinterpret documentation or produce subtly incorrect code. Always review generated code before running it on a production server, and use the thumbs-down button to report inaccurate responses.

Each model has a cost multiplier that reflects how expensive it is to run relative to others. Credits are billed based on the real provider API cost (converted from USD to EUR), then multiplied by this factor before being deducted from your quota. For example, a ×3.0 model costs three times as many credits per message as a ×1.0 model for the same conversation length. Fast models typically have multipliers below ×1; smart and reasoning models are higher. The multiplier for each model is shown in the model picker.

Every message you send consumes credits — proportional to the number of tokens processed (your prompt + conversation history + documentation context, plus the AI's reply), the model's real API cost, and the model's cost multiplier. The circular progress indicator in the chat input area shows how much of your quota remains. When the circle is full you have your full quota available; it shrinks as you consume credits.

Free tier quotas reset weekly. Premium and Platinum tier quotas reset monthly. The exact next reset time is shown in the tooltip when you hover the progress circle in the chat.

DiSkribe has three tiers:

• Free — weekly credit budget, fast model only, no file attachments.

• Premium — monthly credit budget, access to smart models, file attachment support, higher rate limit. Requires a Patreon pledge.

• Platinum — larger monthly credit budget, all models including reasoning, priority support. Requires a higher Patreon pledge.

See the Pricing page for current credit amounts and pledge thresholds.

DiSkribe is community-funded via Patreon. To upgrade:

1. Make a pledge on the DiSkribe Patreon page at the Premium or Platinum threshold.

2. Go to Account → Manage and link your Patreon account.

3. Your tier will be activated automatically once the pledge is verified.

Cancellation takes effect at the next billing cycle. Until then you retain your current tier. After cancellation your account reverts to the Free tier and your remaining premium credit balance is reset. Your chat history is preserved.

No. DiSkribe uses the API endpoints of AI providers (e.g. Google Gemini API). API usage data is explicitly excluded from model training by Google's terms of service for the Gemini API. The same policy applies to any other provider configured in DiSkribe. Your prompts and conversations are not used to improve or fine-tune any AI model.

DiSkribe stores the following data server-side:

• Your Discord identity (ID, username, avatar URL, email).

• Your chat sessions and message history.

• Credit usage per message (token counts and billed credits).

• Files you upload as attachments.

• Feedback and reports you submit.

• Patreon link data (only if you connect your Patreon account).

See the full Privacy Policy for details on retention periods and your rights.

Your prompts and conversation history are sent to the configured AI inference provider (by default Google Gemini) solely to generate a response. This is the essential nature of the service. No other third parties receive your message content. Data is transmitted over TLS and the providers do not retain API data for training purposes.

Yes. Go to Account → Manage → Privacy and request account deletion. This triggers cascading deletion of your profile, chat history, attachments, token records, and all associated data within 30 days, in accordance with GDPR Art. 17.

DiSkribe is hosted in the European Union (France). Data is stored in a local SQLite database on the server. OAuth tokens for third-party services are stored server-side and are never exposed to the browser. All connections use TLS encryption in transit.

DiSkribe supports text-based files such as .sk (Skript), .log, .yml, .yaml, .json, and plain text files. Binary files (images, executables, archives) are not supported. The maximum file size per upload is determined by the administrator. File attachments are only available on Premium and Platinum plans.

For large files, DiSkribe offers a "chunked" reading mode. Instead of injecting the entire file into the AI context (which is expensive in tokens), the AI uses a read_attachment tool to fetch sections of the file on demand. This keeps token usage low while still allowing the AI to analyse large scripts line by line. You can toggle chunked mode per attachment using the switch on the attachment chip.

Uploaded attachments are stored for the duration of the chat session. When you delete a session, all its attachments are removed. When you delete your account, all attachments are permanently deleted. You can also remove individual attachments at any time using the × button on the attachment chip.

Use the thumbs-down button below the message to open the report dialog. You can select a reason (wrong information, outdated content, harmful content, etc.) and add a note. Reports are reviewed to improve the system prompt and documentation grounding.

Credit consumption depends on several factors: the length of your conversation history (all previous messages are re-sent on every turn), the number of documentation pages fetched by the AI tools, any file attachments, and the credit multiplier of the selected model. Switching to a faster / cheaper model and starting a fresh session (to reduce history size) are the most effective ways to reduce consumption.

For account or billing issues, email us at [email protected]. For bugs or feature requests, please open an issue on the project's GitHub repository. Community support is also available in the DiSky Discord server.