AGENTS.md — MTG RFID Companion (ESP-IDF)

Context for AI Agents & Assistant Tools:
This file outlines the architecture, coding guidelines, dependencies, hardware constraints, and build procedures for the MTG RFID Companion project. Read this completely before generating code, refactoring, or suggesting dependencies.

1. Project Overview

The MTG RFID Companion is a hand-held hardware utility for Magic: The Gathering players. It uses an ESP32 microcontroller running ESP-IDF v5.x (FreeRTOS) to scan RFID/NFC tags embedded in card sleeves, query card data locally or via Scryfall's API, render data to a 2.0" ST7789 TFT LCD (240x320), and host a local registration web server.

2. Target Hardware & Specifications

MCU: ESP32-WROOM-32 (Dual-Core Xtensa LX6, 240MHz, 520KB SRAM, 4MB Flash). No external PSRAM.
Display: 2.0" ST7789 SPI TFT LCD (240x320 resolution, RGB565 format). Full-screen JPEG wallpapers supported via tjpgd decoder.
RFID Reader: MFRC522 (13.56 MHz, SPI Interface, ISO/IEC 14443A).
Target Tags: NTAG213 / NTAG215 micro FPC stickers or standard MIFARE ISO cards (7-byte or 4-byte UIDs).
User Input: EC11 Rotary Encoder with integrated push button.
Storage: On-chip SPI Flash partitioned with LittleFS (mounted at /spiffs, size ~1.5 MB).

3. Peripheral Wiring & GPIO Pinout

CRITICAL: Do NOT modify GPIO pin assignments in code generation unless explicitly instructed. SPI bus is shared between ST7789 and MFRC522.

Peripheral	Functional Pin	ESP32 GPIO Pin	Shared Bus / Notes
Shared SPI	SCK / MOSI / MISO	GPIO 18 / 23 / 19	Shared SPI2_HOST (VSPI)
MFRC522 RFID	CS (SDA) / RST	GPIO 5 / GPIO 22	Dedicated CS & Reset
ST7789 Display	CS / DC / RST	GPIO 15 / GPIO 2 / GPIO 4	Dedicated CS, Data/Cmd & Reset
Rotary Encoder	CLK / DT / SW	GPIO 32 / GPIO 33 / GPIO 25	Active low / Interrupt driven
Indicators	Neopixel (WS2812B)	GPIO 27	RMT driver
Audio	Piezo Buzzer	GPIO 14	LEDC PWM channel

4. Software Architecture & FreeRTOS Task Allocation

The project strictly follows a multithreaded, event-driven model using FreeRTOS tasks and the ESP Event Loop.

Core Core Split

Core 0 (System, I/O & Network):
- rc522_task: Continuously polls RFID reader, emits events upon card detection.
- net_task: Manages Wi-Fi reconnects, Scryfall HTTPS fetches (esp_http_client), and esp_http_server.
- storage_task: Reads/writes JSON files to /spiffs via VFS.
Core 1 (User Interface & Rendering):
- ui_task: Handles ST7789 display drawing using esp_lcd and rotary encoder inputs. Maintains UI frame rates without blocking Core 0 I/O.

System State Machine

State transitions must be managed cleanly using FreeRTOS Event Groups or Queue messaging:

STATE_BOOT: Hardware setup, LittleFS mount, Wi-Fi auto-connect.
STATE_SCANNER: Waiting for RFID tag read.
STATE_CARD_VIEW: Displaying fetched/cached card attributes (P/T, Mana, Oracle text).
STATE_LIFE_COUNTER: Interactive 20/40 life counter utilizing rotary encoder.
STATE_REGISTRATION: Local SoftAP mode (192.168.4.1) active to pair unknown UIDs with card names via web browser.

5. Storage Structure (LittleFS `/spiffs`)

Do NOT assume an SD card is attached. All local database operations use standard C file I/O (fopen, fread, fwrite) over LittleFS.

/spiffs/mappings.json — Maps scanned UIDs to Card Names:

{
  "04A2B3C4": "Sol Ring",
  "04E1F2A3": "Atraxa, Praetors' Voice"
}

/spiffs/cards/[UID].json — Cached Scryfall card attribute details:
  JSON

  {
    "name": "Sol Ring",
    "mana_cost": "{1}",
    "type": "Artifact",
    "oracle": "Tap: Add CC.",
    "price_usd": "1.25"
  }

Coding Standards & Conventions

When generating or editing code for this repository:

Framework: Pure ESP-IDF v5.x (C99/C11 standard). Do NOT use Arduino.h headers, functions, or Arduino libraries (Serial.print, delay(), Wire, etc.).

Error Handling: Always wrap ESP-IDF API functions in ESP_ERROR_CHECK() or handle errors gracefully using esp_err_t ret. Log errors via ESP_LOGE().

Memory Management: SRAM is tight (~520KB total, no PSRAM).

    Do NOT allocate huge static buffers on the FreeRTOS task stack.

    Allocate HTTP response buffers and cJSON trees dynamically on the heap (heap_caps_malloc(..., MALLOC_CAP_8BIT)), and always free them immediately after use.

JSON Parsing: Use native cJSON. Check every pointer return for NULL before accessing object fields.

Logging Tag: Define a static tag per source file:
C

static const char *TAG = "MODULE_NAME";

Project Dependencies & Configuration

Managed via idf_component.yml:

abobija/rc522: SPI RFID reader driver.

joltwallet/littlefs: LittleFS VFS port for ESP-IDF.

espressif/esp_lcd: Official display subsystem (ST7789 controller driver).

Build Commands

Build Project: idf.py build

Flash to Device: idf.py -p <PORT> flash

Monitor Logs: idf.py -p <PORT> monitor

Partition Table: Uses custom partitions.csv (Flash allocation: 2MB App, 1.5MB LittleFS data).

Guidance for AI Agents Generating Code

When asked to build features: Write full, compile-ready C files (.c) and header files (.h) targeting ESP-IDF v5.x.

When modifying hardware code: Cross-check the shared SPI bus configurations to prevent SPI2_HOST bus conflicts between ST7789 and MFRC522.

When implementing network code: Use non-blocking esp_http_client requests running on a dedicated FreeRTOS task to keep screen rendering smooth.

5.6 KiB Raw Blame History