visionclaw 0.1.173-beta.0 → 0.1.173

package/CHANGELOG.md

@@ -1,3 +1,15 @@
+## [0.1.173](https://github.com/babelcloud/visionclaw/compare/v0.1.173-beta.0...v0.1.173) (2026-03-29)
+
+
+### Features
+
+* add default search skill with Google Lens image identification ([627857a](https://github.com/babelcloud/visionclaw/commit/627857ae8f6181f09d30742daa241f3d00e3d156))
+* add machine management service ([1d94b0f](https://github.com/babelcloud/visionclaw/commit/1d94b0f31a75e71e088cae9412b9b44086454d88))
+* enhance delete machine confirmation prompt ([21b477b](https://github.com/babelcloud/visionclaw/commit/21b477b3b2d45a9ce20b1597d8b9f327f51ac130))
+* enhance machine management service with authentication and structured location support ([262136a](https://github.com/babelcloud/visionclaw/commit/262136af944db7c69b1f25d1b2a32a10b607515f))
+
+
+
 ## [0.1.172](https://github.com/babelcloud/visionclaw/compare/v0.1.171...v0.1.172) (2026-03-28)
 
 

package/dist/builtin-skills/search/SKILL.md

@@ -0,0 +1,66 @@
+---
+description: Use this skill for web search using SerpAPI. Covers general search, image/artwork identification via Google Lens, and other specialized search engines.
+---
+
+# Web Search with SerpAPI
+
+You have access to SerpAPI via the `mcp__serpapi__search` tool. It supports many engines — use the right one for the task.
+
+## General Web Search
+
+```json
+{"params": {"q": "your query", "engine": "google_light"}, "mode": "compact"}
+```
+
+Use `google_light` for most searches (faster, cheaper). Use `google` when you need full results including knowledge graph, AI overview, etc.
+
+## Image / Artwork Identification (Google Lens)
+
+To identify what's in an image (artwork, object, landmark, product, etc.), use a **two-step** Google Lens flow. This is much more effective than `google_reverse_image`, which only returns visually similar images without identifying the subject.
+
+### Step 1: Google Lens lookup
+
+```json
+{"params": {"engine": "google_lens", "url": "<public_image_url>"}, "mode": "compact"}
+```
+
+- The image URL **must be publicly accessible**. If you only have a local file, upload it first (e.g. to catbox.moe via `curl -F "reqtype=fileupload" -F "fileToUpload=@/path/to/image.jpg" https://catbox.moe/user/api.php`).
+- The response includes `visual_matches` and an `ai_overview` object.
+- Extract `ai_overview.page_token` from the response.
+
+### Step 2: Get full AI Overview
+
+```json
+{"params": {"engine": "google_ai_overview", "page_token": "<token_from_step_1>"}, "mode": "compact"}
+```
+
+This returns the full AI Overview text with detailed identification — artist, title, date, medium, provenance, and context for artworks; species, model, landmark name, etc. for other subjects.
+
+### Example: Identifying a painting
+
+```
+# Step 1
+search(params={"engine": "google_lens", "url": "https://example.com/mystery-painting.jpg"}, mode="compact")
+# Response includes: ai_overview.page_token = "abc123..."
+
+# Step 2
+search(params={"engine": "google_ai_overview", "page_token": "abc123..."}, mode="compact")
+# Response: Full identification with artist, title, year, etc.
+```
+
+## Other Useful Engines
+
+| Engine | Use case |
+|--------|----------|
+| `google_light` | General web search (default) |
+| `google` | Full Google search with knowledge graph |
+| `google_lens` | Image identification (step 1) |
+| `google_ai_overview` | AI overview details (step 2) |
+| `google_news` | News search |
+| `google_images` | Image search by text query |
+| `google_shopping` | Product search and price comparison |
+| `google_flights` | Flight search |
+| `google_hotels` | Hotel search |
+| `youtube_search` | YouTube video search |
+| `google_local` | Local business search |
+| `google_jobs` | Job search |

package/dist-agent/bundle.cjs

@@ -20333,7 +20333,7 @@ var require_prompts3 = __commonJS({
 
 // dist/utils/version-check.js
 function isBundled() {
-  const v10 = "0.1.173-beta.0";
+  const v10 = "0.1.173";
   return typeof v10 === "string" && v10 !== "undefined";
 }
 function getPackageRoot() {
@@ -20350,7 +20350,7 @@ function getInstallationInfo() {
   };
 }
 function getCurrentVersion() {
-  const bundledVersion = "0.1.173-beta.0";
+  const bundledVersion = "0.1.173";
   if (bundledVersion && bundledVersion !== "undefined") {
     return bundledVersion;
   }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "visionclaw",
-  "version": "0.1.173-beta.0",
+  "version": "0.1.173",
   "description": "A personal assistant agent that runs on your desktop, receives commands from messaging channels, and executes tasks autonomously.",
   "type": "module",
   "main": "dist/index.js",