feat(providers): add Evolink image generation provider

[EvoLinkAI]

Description

Add a dedicated Evolink provider for image generation via EvoLink.ai API.

EvoLink's image API uses an async submit/poll pattern:

• Submit: POST https://api.evolink.ai/v1/images/generations → returns task id
• Poll: GET https://api.evolink.ai/v1/tasks/{task_id} → status + results[]

This pattern differs from ClawMate's existing providers (OpenAI-compatible sync, Volcengine ARK sync, Gemini SDK sync), so a dedicated provider is needed — the http-async provider's hardcoded body format and status values are incompatible with EvoLink's API.

Changes (2 files, +329)

File	Change
src/core/providers/evolink.ts (new)	Dedicated async submit/poll provider
src/core/providers/registry.ts	Register evolink type + auto-detection via api.evolink.ai base URL

Provider Features

• Async task submission with configurable polling interval/timeout
• Configurable model, size, quality parameters
• Supports EVOLINK_API_KEY environment variable fallback
• Auto-detection from baseUrl containing api.evolink.ai
• Graceful degradation: missing API key creates unavailable provider (no crash)

Available Models

EvoLink offers 20+ image models via a unified API: GPT Image (1.5/2), Midjourney V7, Nano Banana (Beta/Pro), Qwen Image Edit, Seedream (4.0/4.5/5.0 Lite), WAN 2.5, Z Image Turbo, and more.

Config Example

{
  "defaultProvider": "evolink",
  "providers": {
    "evolink": {
      "type": "evolink",
      "apiKey": "sk-...",
      "model": "gpt-image-1.5",
      "size": "1024x1024",
      "quality": "high"
    }
  }
}

Testing

• ✅ Provider loads and registers successfully
• ✅ Auto-detection via baseUrl heuristic works
• ✅ Missing API key creates unavailable provider (no crash)
• ✅ EVOLINK_API_KEY env var fallback works
• ✅ Manifest consistency check (npm run clawmate:plugin:check) passes
• ✅ HTTP request format verified (server responds with proper error, not malformed request)

API Documentation

• EvoLink docs: https://docs.evolink.ai
• Base URL: https://api.evolink.ai
• Auth: Authorization: Bearer ${EVOLINK_API_KEY}
• Key registration: https://evolink.ai/dashboard/keys

---

🤖 Generated with Claude Code

[EvoLinkAI]

✅ 真实验证通过

用真实 EvoLink API key 跑通了完整的端到端生图流程：

curl 直接验证

POST https://api.evolink.ai/v1/images/generations → 200 (task created)
GET  https://api.evolink.ai/v1/tasks/{task_id}   → 200 (polling)

结果：19 次轮询后 status: completed，图片 URL 正常返回。

Provider 代码端到端

createEvolinkProvider({ model: 'gpt-image-1.5', size: '1024x1024', quality: 'medium' })
  → generate({ prompt: 'A fluffy white cat...' })
  → ✅ SUCCESS (39.8s)

生成图片：

本地校验汇总

检查项	状态
Provider 加载注册	✅
baseUrl 自动检测	✅
缺失 API key → unavailable	✅
EVOLINK_API_KEY 环境变量回退	✅
异步 submit/poll 轮询	✅
status: processing (进行中)	✅
status: completed (成功)	✅
npm run clawmate:plugin:check	✅
真实 API 端到端	✅