AgentSkillsCN

input-abstraction-layer

为电视遥控器和游戏手柄打造统一的输入层。在对三星、LG、海信、Xbox 和 PS5 环境中的按键码、长按操作、重复输入行为，以及“返回”/“菜单”语义进行规范化处理时使用。

SKILL.md
--- frontmatter
name: input-abstraction-layer
description: Build a unified input layer for TV remotes and gamepads. Use when normalizing key codes, long-press, repeat behavior, and Back/Menu semantics across Samsung, LG, Hisense, Xbox, and PS5 environments.

Input Abstraction Layer

Overview

Normalize heterogeneous platform key events into one app-level action contract. Treat platform keycodes as adapter concerns, not feature code concerns.

Core Contract

Expose canonical actions:

  • •NAV_UP, NAV_DOWN, NAV_LEFT, NAV_RIGHT
  • •SELECT, BACK, MENU, PLAY_PAUSE
  • •SEEK_FWD, SEEK_BACK, PAGE_UP, PAGE_DOWN

Workflow

  1. •Create per-platform adapters that map native events to canonical actions.
  2. •Centralize repeat-rate and long-press policy in one dispatcher.
  3. •Emit input source metadata (remote, gamepad, keyboard) for telemetry.
  4. •Gate feature-level handlers behind canonical actions only.
  5. •Add platform fallback paths for missing keys.

Reliability Rules

  • •Debounce duplicate native key bursts before dispatch.
  • •Ensure Back is idempotent for fast repeat presses.
  • •Disable key-repeat on destructive actions (sign-out, purchase confirm).
  • •Log unknown keycodes with platform/build context.

Validation Checklist

  • •Same user intent triggers same action on all targets.
  • •Long-press thresholds are consistent and measurable.
  • •No double-dispatch during key repeat storms.
  • •Back/Menu precedence is consistent with product rules.

References

  • •Use references/keymap-matrix.md as the baseline mapping matrix.