MQL5-SKILLS

A collection of reusable MQL5 skills for building, testing, and automating trading strategies.

Repository Structure

MQL5-SKILLS/
|-- LICENSE
|-- README.md
`-- skills/
    |-- candles-and-series/
    |   `-- SKILL.md
    |-- captions-davinci/
    |   |-- SKILL.md
    |   |-- style.json
    |   |-- agents/
    |   |   `-- openai.yaml
    |   |-- examples/
    |   |   `-- clip-config.json
    |   |-- fonts/
    |   |   |-- OFL.txt
    |   |   |-- Poppins-Bold.ttf
    |   |   `-- README.md
    |   `-- scripts/
    |       `-- captions.py
    |-- compile-mql5/
    |   `-- SKILL.md
    |-- edit-davinci/
    |   |-- SKILL.md
    |   |-- agents/
    |   |   `-- openai.yaml
    |   `-- scripts/
    |       `-- edit_tools.py
    |-- expert-advisors/
    |   `-- SKILL.md
    |-- git/
    |   `-- SKILL.md
    |-- paint-objects/
    |   `-- SKILL.md
    `-- run-backtests/
        `-- SKILL.md

Skills Overview

• candles-and-series: Conventions for loading and indexing candle data in MQL5, especially when using series mode and recent-candle-first logic.
• captions-davinci: Animated burned-in captions for vertical shorts from a word-level transcript, with the exact style stored in style.json so it can be reproduced identically.
• compile-mql5: Workflow for compiling MQL5 experts, indicators, and scripts with MetaEditor and checking the real compilation log.
• edit-davinci: Basic DaVinci Resolve editing through the MCP: connecting via the bridge, project setup, cutting clips into a timeline, rendering, and reviewing the render with extracted frames.
• expert-advisors: Conventions for structuring MQL5 expert advisors, especially grouped inputs and practical parameter comments.
• git: Git conventions for this repo, including branch naming, commit message format, and issue references.
• paint-objects: Rules for creating and updating chart objects in MQL5, including anchoring, repaint-safe updates, and redraw behavior.
• run-backtests: Guided workflow for configuring and launching MetaTrader 5 backtests with confirmed inputs and real result validation.