@anionex/dsh-vision-toolkit 0.1.5 → 0.1.6

package/src/config.ts

@@ -15,15 +15,24 @@ import { VisionToolkitError } from './errors.ts'
 /** Settings document namespace owned by this plugin. */
 export const VISION_TOOLKIT_SETTINGS_NAMESPACE = settingsNamespace('vision-toolkit')
 
+/** Browser-compatible default shared with the vendored Python client. */
+export const DEFAULT_VISION_USER_AGENT = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/126.0.0.0 Safari/537.36'
+
 /** Full user-facing configuration; every field defaults at the schema boundary. */
 export interface VisionToolkitConfig {
   provider?: {
-    /** OpenAI-compatible chat/completions base URL. */
+    /** Provider API base URL. */
     baseUrl?: string
     /** DSH Credential reference holding the API key (an environment-style name). */
     credential?: string
     /** Multimodal model name. */
     model?: string
+    /** Vision request protocol: OpenAI Chat Completions or Anthropic Messages. */
+    protocol?: 'openai' | 'anthropic'
+    /** Anthropic thinking field behavior; `omit` leaves model defaults untouched. */
+    anthropicThinking?: 'omit' | 'disabled' | 'adaptive'
+    /** Outbound User-Agent for provider requests and connection tests. */
+    userAgent?: string
   }
   /** Vision output language (`zh` or `en`). */
   language?: 'zh' | 'en'
@@ -53,6 +62,9 @@ export const Config: Schema<VisionToolkitConfig> = z.object({
     baseUrl: z.string().default('https://api.inferera.com/v1'),
     credential: z.string().default('VISION_API_KEY'),
     model: z.string().default('gemini-3.6-flash'),
+    protocol: z.union(['openai', 'anthropic'] as const).default('openai'),
+    anthropicThinking: z.union(['omit', 'disabled', 'adaptive'] as const).default('omit'),
+    userAgent: z.string().default(DEFAULT_VISION_USER_AGENT),
   }),
   language: z.union(['zh', 'en'] as const).default('zh'),
   timeoutMs: z.number().default(60000),
@@ -73,6 +85,9 @@ export interface ResolvedVisionToolkitConfig {
     baseUrl: string
     credential: CredentialRef
     model: string
+    protocol: 'openai' | 'anthropic'
+    anthropicThinking: 'omit' | 'disabled' | 'adaptive'
+    userAgent: string
   }
   language: 'zh' | 'en'
   timeoutMs: number
@@ -121,6 +136,18 @@ export function resolveConfig(config: VisionToolkitConfig = {}): ResolvedVisionT
   if (model.length === 0) {
     throw new VisionToolkitError('config', 'provider.model must not be empty')
   }
+  const protocol = provider.protocol ?? 'openai'
+  if (protocol !== 'openai' && protocol !== 'anthropic') {
+    throw new VisionToolkitError('config', 'provider.protocol must be "openai" or "anthropic"')
+  }
+  const anthropicThinking = provider.anthropicThinking ?? 'omit'
+  if (anthropicThinking !== 'omit' && anthropicThinking !== 'disabled' && anthropicThinking !== 'adaptive') {
+    throw new VisionToolkitError('config', 'provider.anthropicThinking must be "omit", "disabled", or "adaptive"')
+  }
+  const userAgent = (provider.userAgent ?? DEFAULT_VISION_USER_AGENT).trim()
+  if (userAgent.length === 0) {
+    throw new VisionToolkitError('config', 'provider.userAgent must not be empty')
+  }
   const language = config.language ?? 'zh'
   if (language !== 'zh' && language !== 'en') {
     throw new VisionToolkitError('config', 'language must be "zh" or "en"')
@@ -161,7 +188,7 @@ export function resolveConfig(config: VisionToolkitConfig = {}): ResolvedVisionT
   }
   const allowedDirs = (config.allowedDirs ?? []).map(dir => dir.trim()).filter(dir => dir.length > 0)
   return {
-    provider: { baseUrl, credential, model },
+    provider: { baseUrl, credential, model, protocol, anthropicThinking, userAgent },
     language,
     timeoutMs,
     maxImageBytes,

package/src/runtime.ts

@@ -768,6 +768,9 @@ export class VisionToolkitRuntime {
       VISION_API_KEY: resolved.value,
       VISION_BASE_URL: this.config.provider.baseUrl,
       VISION_MODEL: this.config.provider.model,
+      VISION_API_PROTOCOL: this.config.provider.protocol === 'anthropic' ? 'anthropic' : 'chat_completions',
+      VISION_ANTHROPIC_THINKING: this.config.provider.anthropicThinking,
+      VISION_USER_AGENT: this.config.provider.userAgent,
       LANG: this.config.language,
     }
   }
@@ -832,6 +835,9 @@ export class VisionToolkitRuntime {
       provider: {
         baseUrl: env.VISION_BASE_URL,
         model: env.VISION_MODEL,
+        protocol: env.VISION_API_PROTOCOL,
+        anthropicThinking: env.VISION_ANTHROPIC_THINKING,
+        userAgent: env.VISION_USER_AGENT,
         language: env.LANG,
         credentialSha256: createHash('sha256').update(env.VISION_API_KEY).digest('hex'),
       },
@@ -1735,9 +1741,19 @@ export class VisionToolkitRuntime {
           const endpoint = `${this.config.provider.baseUrl}/models`
           try {
             const started = Date.now()
+            const headers: Record<string, string> = {
+              Accept: 'application/json',
+              'User-Agent': this.config.provider.userAgent,
+            }
+            if (this.config.provider.protocol === 'anthropic') {
+              headers['x-api-key'] = resolvedCredential.value
+              headers['anthropic-version'] = '2023-06-01'
+            } else {
+              headers.Authorization = `Bearer ${resolvedCredential.value}`
+            }
             const response = await fetch(endpoint, {
               method: 'GET',
-              headers: { Authorization: `Bearer ${resolvedCredential.value}`, Accept: 'application/json' },
+              headers,
               signal: operation.signal,
             })
             operation.metrics.upstreamMs += Date.now() - started

package/src/upstream.ts

@@ -38,6 +38,9 @@ export interface UpstreamEnvironment {
   VISION_API_KEY: string
   VISION_BASE_URL: string
   VISION_MODEL: string
+  VISION_API_PROTOCOL: 'chat_completions' | 'anthropic'
+  VISION_ANTHROPIC_THINKING: 'omit' | 'disabled' | 'adaptive'
+  VISION_USER_AGENT: string
   LANG: 'zh' | 'en'
 }
 
@@ -612,6 +615,9 @@ export class UpstreamAdapter {
           VISION_API_KEY: options.env.VISION_API_KEY,
           VISION_BASE_URL: options.env.VISION_BASE_URL,
           VISION_MODEL: options.env.VISION_MODEL,
+          VISION_API_PROTOCOL: options.env.VISION_API_PROTOCOL,
+          VISION_ANTHROPIC_THINKING: options.env.VISION_ANTHROPIC_THINKING,
+          VISION_USER_AGENT: options.env.VISION_USER_AGENT,
           LANG: options.env.LANG,
           VISION_ENV_FILE: join(prepared.cleanHome, 'vision.env'),
         }),

package/src/web.ts

@@ -289,7 +289,7 @@ export class VisionToolkitWebBackend {
 }
 
 /**
- * Attach optional Web routes whenever an webServer service is present.
+ * Attach optional Web routes whenever a webServer service is present.
  * @param ctx - plugin context owning route effects.
  * @param backend - Settings handler.
  * @param artifacts - signed Artifact handler.

package/vendor/agent-vision-toolkit/CHANGELOG.md

@@ -2,6 +2,19 @@
 
 All notable user-facing changes to agent-vision-toolkit are documented in this file.
 
+## [Unreleased]
+
+### Added
+
+- Let the shared Python vision client call either Chat Completions or Responses APIs, including optional reasoning effort and explicit `store: false` data handling.
+- Add native Anthropic Messages requests with protocol-specific authentication, image sources, optional thinking control, and text-block response extraction.
+- Rewrite OpenAI Chat Completions `image_url` blocks through the existing vision-description pipeline with a host-neutral channel note.
+
+### Fixed
+
+- Send a browser-compatible, configurable User-Agent from the shared Python vision client so Cloudflare-backed OpenAI-compatible endpoints do not reject the default `Python-urllib` signature.
+- Honor `Retry-After` and retry Anthropic 529 overload responses.
+
 ## [0.1.0] - 2026-08-07
 
 ### Added

package/vendor/agent-vision-toolkit/README.md

@@ -6,6 +6,7 @@
 
 # agent-vision-toolkit
 
+[![X (Twitter)](https://img.shields.io/badge/-@anion__ex-000000?style=flat-square&logo=x&logoColor=white)](https://x.com/anion_ex)
 [![GitHub stars](https://img.shields.io/github/stars/Anionex/agent-vision-toolkit?style=flat-square&logo=github)](https://github.com/Anionex/agent-vision-toolkit/stargazers)
 [![GitHub forks](https://img.shields.io/github/forks/Anionex/agent-vision-toolkit?style=flat-square&logo=github)](https://github.com/Anionex/agent-vision-toolkit/forks)
 [![License: MIT](https://img.shields.io/github/license/Anionex/agent-vision-toolkit?style=flat-square&color=4EAA25)](https://github.com/Anionex/agent-vision-toolkit/blob/main/LICENSE)
@@ -136,7 +137,7 @@ When to use them, the order in which to call tools, and how to verify the result
 
 > Read https://github.com/Anionex/agent-vision-toolkit/blob/main/AGENT_INSTALL.md in full, then install the appropriate vision proxy or native extension/plugin for the agent application we are currently using. If the vision API is not configured, locate the configuration file for the current operating system and guide me through setting `VISION_API_KEY`, `VISION_BASE_URL`, and `VISION_MODEL`.
 
-All you need to prepare is an OpenAI-compatible multimodal API base URL, API key, and model name. The agent will guide you through writing them to the appropriate configuration file.
+All you need to prepare is a multimodal API supporting OpenAI Chat Completions, OpenAI Responses, or Anthropic Messages, plus its base URL, API key, and model name. The agent will guide you through writing them to the appropriate configuration file.
 
 > After installing the optional integration and restarting the agent, paste an image directly or let the model call its built-in image tool. Pi, Oh My Pi, and OpenCode use single-file [native extensions](extensions/) rather than the proxy; see each agent's documentation.
 
@@ -151,7 +152,7 @@ VISION_BASE_URL=https://openrouter.ai/api/v1
 VISION_MODEL=google/gemini-3.6-flash
 ```
 
-Any OpenAI-compatible endpoint that supports `/chat/completions` with `image_url` works (e.g. Aliyun DashScope: `https://dashscope.aliyuncs.com/compatible-mode/v1` + `qwen-vl-max-latest`). Add `LANG=en` for English descriptions (default is Chinese).
+Any OpenAI-compatible endpoint that supports `/chat/completions` with `image_url` works (e.g. Aliyun DashScope: `https://dashscope.aliyuncs.com/compatible-mode/v1` + `qwen-vl-max-latest`). The Python client/proxy can also use `/responses` with `input_image` by setting `VISION_API_PROTOCOL=responses`, or Anthropic Messages by setting `VISION_API_PROTOCOL=anthropic` and a base URL ending in `/v1` (not `/messages`). Add `LANG=en` for English descriptions (default is Chinese).
 
 **2. Put the CLIs on your PATH:**
 
@@ -322,14 +323,18 @@ Codex -> 127.0.0.1:19100 -> your existing text-only upstream
 <details>
 <summary><b>Environment variables</b></summary>
 
-The toolkit and proxy use only these environment variables; just three are required:
+The standalone CLIs and Python proxy use these environment variables; just three are required. The native Pi and OpenCode extensions use their own settings and currently call `/chat/completions` only.
 
 | Variable | Required | Description |
 |---|---:|---|
 | `VISION_API_KEY` | Yes | API key of the multimodal model |
-| `VISION_BASE_URL` | Yes | OpenAI-compatible API base URL |
+| `VISION_BASE_URL` | Yes | Provider API base URL; include `/v1` but not the protocol endpoint such as `/messages` |
 | `VISION_MODEL` | Yes | Multimodal model name |
 | `LANG` | No | Vision model output language: `zh` (Chinese) or `en` (English); default `zh` |
+| `VISION_API_PROTOCOL` | No | Python client/proxy protocol: `chat_completions` (default), `responses`, or `anthropic`; Anthropic mode uses `x-api-key` and `anthropic-version` |
+| `VISION_REASONING_EFFORT` | No | Optional provider-supported reasoning effort for the Python client/proxy when using `responses` |
+| `VISION_ANTHROPIC_THINKING` | No | Anthropic thinking mode. `omit` (default) sends no thinking field and has the broadest compatibility. Use `disabled` or `adaptive` only when the selected model documents that mode; restore `omit` first if the provider returns HTTP 400. Manual `enabled` plus `budget_tokens` is not exposed. |
+| `VISION_USER_AGENT` | No | Outbound User-Agent for the Python client/proxy; defaults to a browser-compatible value and can be overridden for provider requirements |
 
 </details>
 
@@ -348,7 +353,7 @@ The route whose connection (TCP/TLS handshake) succeeds is kept in memory and re
 ## Prerequisites
 
 - A coding agent already working with a model, including a text-only model such as DeepSeek V4
-- An OpenAI-compatible vision API that supports `/chat/completions` and `image_url`
+- A vision API supporting OpenAI Chat Completions, OpenAI Responses, or Anthropic Messages; select the latter two with `VISION_API_PROTOCOL=responses` or `VISION_API_PROTOCOL=anthropic`
 - No other configuration is required
 
 ## FAQ
@@ -396,4 +401,4 @@ To reduce costs further, you can use a locally deployed small multimodal side mo
 
 If agent-vision-toolkit saves you time, you are welcome to star it, share it, contribute, or [sponsor the project](FUNDING.md).
 
-I'm [anionex](https://anionex.me/), an AI-native developer who once ranked No. 4 on GitHub's global developer trending list, with more than 16k stars across my projects. If you would like to follow my future work, [follow me on GitHub](https://github.com/Anionex).
+I'm [anionex](https://anionex.me/), an AI-native developer who once ranked No. 4 on GitHub's global developer trending list, with more than 16k stars across my projects. If you would like to follow my future work, [follow me on X](https://x.com/anion_ex) or [GitHub](https://github.com/Anionex).

package/vendor/agent-vision-toolkit/UPSTREAM_MANIFEST.json

@@ -1,14 +1,14 @@
 {
   "schemaVersion": 1,
   "repository": "https://github.com/Anionex/agent-vision-toolkit",
-  "version": "v0.1.0+snapshot.8ccf9d7",
-  "commit": "8ccf9d7d89bc395ad7dd19dde840401a5b84623d",
-  "contentSha256": "99af16244fef18745ab0842ff555a54d1829d4ed588adfa7e05756beffe813c2",
+  "version": "v0.1.0+snapshot.bc9803d",
+  "commit": "bc9803d7d6300c864d17460ecbb33540b26638e0",
+  "contentSha256": "6936f8fba3ed40f54267245e717603c2622649ce4bd7ccf3c1b23f39047b1f3f",
   "files": [
     {
       "path": "CHANGELOG.md",
-      "bytes": 913,
-      "sha256": "3eb86d51d5fd89714045e73a29ebd4aa5b146e9bf5f8c8620cc9b452a8d4bbe0"
+      "bytes": 1671,
+      "sha256": "0bb23408797c8e5389abc0b34d47b43907316f8bd106420e4c05f25606bccde1"
     },
     {
       "path": "LICENSE",
@@ -17,8 +17,8 @@
     },
     {
       "path": "README.md",
-      "bytes": 22087,
-      "sha256": "c530e6d123a2a535742ddf1637ca78150b8f96cd2488d5763fc0220ba75aab05"
+      "bytes": 23625,
+      "sha256": "55d880175b51d288f5b817d064ebb79e6c8663ee0bb1817f4a0390ab18b5c3ee"
     },
     {
       "path": "bin/crop",
@@ -80,10 +80,15 @@
       "bytes": 3978,
       "sha256": "70e04fea9bf452f35d12bed0c19c7e2be46934a5423caa778b24da82b78f2251"
     },
+    {
+      "path": "tests/test_vision_client.py",
+      "bytes": 17933,
+      "sha256": "eb31118751a5724385f486b98d45af83cad1cf7e4506b79ddf9a840d08f22367"
+    },
     {
       "path": "vision_client.py",
-      "bytes": 6117,
-      "sha256": "5450985379a0c1e6460594a8739966fe1e2489311ce473b1d92021ccf41d0be8"
+      "bytes": 10610,
+      "sha256": "7a899d08b1c65721c9fdfd42d9d3288671c149fdc919b5dedb5b36cc6b06ddce"
     }
   ]
 }