@floatingsk/dsh-vision 0.1.5 → 0.1.6

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -104,15 +104,19 @@ Agent 会自动选择合适的工具(读文字走 OCR,看画面走 VLM)。
104
104
 
105
105
  ### 多供应商 VLM
106
106
 
107
- 内置四家国内供应商,`describe_image` 可传 `provider` 参数选择(留空用 `defaultProvider`):
107
+ 内置五家供应商(默认 DeepSeek 原生视觉,最便宜),`describe_image` 可传 `provider` 参数选择(留空用 `defaultProvider`):
108
108
 
109
109
  ```yaml
110
110
  - insert:
111
111
  - id: dsh-vision
112
112
  name: 'dsh-vision'
113
113
  config:
114
- defaultProvider: 'bailian' # 默认供应商
114
+ defaultProvider: 'deepseek' # 默认供应商
115
115
  providers:
116
+ deepseek: # DeepSeek 原生视觉(V4-Flash-Vision-Exp,约0.001元/张)
117
+ baseUrl: 'https://api.deepseek.com'
118
+ model: 'deepseek-v4-flash-vision-exp'
119
+ apiKeyEnv: 'DEEPSEEK_API_KEY'
116
120
  bailian: # 阿里云百炼
117
121
  baseUrl: 'https://dashscope.aliyuncs.com/compatible-mode/v1'
118
122
  model: 'qwen3-vl-flash' # 或 qwen3-vl-plus / qwen-vl-ocr
package/lib/core.js CHANGED
@@ -6,7 +6,7 @@ import { homedir } from "node:os";
6
6
  /** VLM 请求超时(毫秒) */
7
7
  export const VLM_TIMEOUT_MS = 120_000;
8
8
  /** VLM 输出 token 上限 */
9
- export const VLM_MAX_TOKENS = 1024;
9
+ export const VLM_MAX_TOKENS = 2048;
10
10
  /** 上传前图片压缩的最长边(像素) */
11
11
  export const VLM_MAX_IMAGE_DIM = 2048;
12
12
  /** 压缩 JPEG 质量(0-100) */
package/lib/index.js CHANGED
@@ -38,9 +38,14 @@ export const Config = z.object({
38
38
  /** 本地 OCR 可执行文件路径(默认指向插件 bin/vision-ocr) */
39
39
  ocrBin: z.string().default(""),
40
40
  /** 默认 VLM 供应商 id(对应 providers 的键名) */
41
- defaultProvider: z.string().default("bailian"),
41
+ defaultProvider: z.string().default("deepseek"),
42
42
  /** 多供应商 VLM 配置:每个供应商为 OpenAI 兼容端点 */
43
43
  providers: z.object({
44
+ deepseek: providerSchema.default({
45
+ baseUrl: "https://api.deepseek.com",
46
+ model: "deepseek-v4-flash-vision-exp",
47
+ apiKeyEnv: "DEEPSEEK_API_KEY",
48
+ }),
44
49
  bailian: providerSchema.default({
45
50
  baseUrl: "https://dashscope.aliyuncs.com/compatible-mode/v1",
46
51
  model: "qwen3-vl-flash",
package/mcp/server.js CHANGED
@@ -34,12 +34,13 @@ const ocrBin = join(__dirname, "..", "bin", "vision-ocr");
34
34
 
35
35
  // ── 内置供应商(与插件 Config 默认值一致) ──────────────────────────────────
36
36
  const PROVIDERS = {
37
+ deepseek: { baseUrl: "https://api.deepseek.com", model: "deepseek-v4-flash-vision-exp", apiKeyEnv: "DEEPSEEK_API_KEY" },
37
38
  bailian: { baseUrl: "https://dashscope.aliyuncs.com/compatible-mode/v1", model: "qwen3-vl-flash", apiKeyEnv: "DASHSCOPE_API_KEY" },
38
39
  siliconflow: { baseUrl: "https://api.siliconflow.cn/v1", model: "Qwen/Qwen2.5-VL-7B-Instruct", apiKeyEnv: "SILICONFLOW_API_KEY" },
39
40
  zhipu: { baseUrl: "https://open.bigmodel.cn/api/paas/v4", model: "glm-4v-flash", apiKeyEnv: "ZHIPU_API_KEY" },
40
41
  volcengine: { baseUrl: "https://ark.cn-beijing.volces.com/api/v3", model: "doubao-seed-1.6-vision", apiKeyEnv: "ARK_API_KEY" },
41
42
  };
42
- const DEFAULT_PROVIDER = "bailian";
43
+ const DEFAULT_PROVIDER = "deepseek";
43
44
 
44
45
  // ── 工具实现 ─────────────────────────────────────────────────────────────────
45
46
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@floatingsk/dsh-vision",
3
- "version": "0.1.5",
3
+ "version": "0.1.6",
4
4
  "description": "为 DeepSeek Harness(dsh 插件)与 Claude Code(MCP)提供图片理解能力:本地 OCR(macOS Vision / Windows 内置 OCR)+ 云端 VLM(多供应商,OpenAI 兼容)",
5
5
  "type": "module",
6
6
  "main": "lib/index.js",