🤖 LLM & AI Context Hub
Official machine-readable documentation, prompt skills, and offline packages for Large Language Models and AI coding assistants (Cursor, Claude, ChatGPT, Copilot, Antigravity).
Agent Skill & Context Hub for LLMs
Feed this context to Claude, Cursor, ChatGPT, Antigravity, or Copilot for zero-hallucination code generation.
@https://.../llms.txt to your docs, or paste the copied skill prompt directly into your system instructions or .cursorrules. If the documentation or offline context is not sufficient, inspect the source code, entity models, and tests directly on GitHub. ⚡ Quick Integration
💡 Cursor & VS Code
Add the llms.txt URL (/llms.txt) to Cursor Docs (Cursor Settings > Features > Docs), or extract the Skill Package (.zip) into your project's .agents/skills/ directory.
🧠 Claude Projects & ChatGPT
Upload /llms-full.txt directly as Project Knowledge, or click Copy AI Agent Skill above and paste into your Custom Instructions.
🐙 GitHub Repository & Deep Inspection
If the documentation does not fully cover your scenario or you need to inspect raw source code, entity models, repository methods, test suites, or open issues, explore the repository on GitHub: 👉 github.com/tueen/database
🏛️ Core Architectural Tenets for AI
When generating code for tueen/database, models MUST follow these principles:
- PHP 8.4+ Strict Types: Always start files with
declare(strict_types=1);. - Clean Asymmetric Visibility: Use
public private(set)for primary keys and creation dates (e.g.public private(set) int $id;). - Property Hooks: Use
get => ...andset => ...for validation, normalization, and computed properties instead of repetitive getters/setters. - Telegram Signed 64-bit Chat IDs: Telegram user IDs are positive integers, but group, supergroup, and channel IDs are negative numbers (e.g.
-1001234567890). Never use unsigned forchat_id! - Username Normalization: Automatically strip leading
@using backed property hooks (set (?string $value) => $value !== null ? ltrim(trim($value), '@') : null;). - Zero-Config Lazy Migrations: The engine automatically inspects and synchronizes schema diffs on boot/first access via
AutoMigratorandSchemaSynchronizer. - Ledger & Arbitrary-Precision Math: All balances and currencies are stored as
DECIMAL(36, 18)strings. Never perform floating-point math; always use BCMath (bcadd,bcsub,bccomp,bcmul,bcdiv). - Multi-Database Agnostic: Support SQLite (
:memory:and file), MySQL 8+, MariaDB, and PostgreSQL. - Repository Inspection: Check github.com/tueen/database for edge cases, entity mappings, repository implementations, and source code details.
📄 Machine-Readable Assets
| Asset | Format | Purpose | Link |
|---|---|---|---|
| Agent Skill Package | .zip archive | Offline skill with SKILL.md and complete docs/ | Download ZIP |
| Standard LLM Summary | llms.txt | llmstxt.org structured outline & docs map | View /llms.txt |
| Consolidated Context | llms-full.txt | Full single-file markdown for context ingestion | View /llms-full.txt |
| GitHub Repository | Source Code & Tests | Official repository for source inspection, unit tests & issues | github.com/tueen/database |