# Send Watcha Finder to your agent
Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.
## Fast path
- Download the package from Yavira.
- Extract it into a folder your agent can access.
- Paste one of the prompts below and point your agent at the extracted folder.
## Suggested prompts
### New install

```text
I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete.
```
### Upgrade existing

```text
I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run.
```
## Machine-readable fields
```json
{
  "schemaVersion": "1.0",
  "item": {
    "slug": "watcha-finder",
    "name": "Watcha Finder",
    "source": "tencent",
    "type": "skill",
    "category": "开发工具",
    "sourceUrl": "https://clawhub.ai/Charipoter/watcha-finder",
    "canonicalUrl": "https://clawhub.ai/Charipoter/watcha-finder",
    "targetPlatform": "OpenClaw"
  },
  "install": {
    "downloadUrl": "/downloads/watcha-finder",
    "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=watcha-finder",
    "sourcePlatform": "tencent",
    "targetPlatform": "OpenClaw",
    "packageFormat": "ZIP package",
    "primaryDoc": "SKILL.md",
    "includedAssets": [
      "SKILL.md",
      "evals/evals.json"
    ],
    "downloadMode": "redirect",
    "sourceHealth": {
      "source": "tencent",
      "status": "healthy",
      "reason": "direct_download_ok",
      "recommendedAction": "download",
      "checkedAt": "2026-05-07T17:22:31.273Z",
      "expiresAt": "2026-05-14T17:22:31.273Z",
      "httpStatus": 200,
      "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=afrexai-annual-report",
      "contentType": "application/zip",
      "probeMethod": "head",
      "details": {
        "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=afrexai-annual-report",
        "contentDisposition": "attachment; filename=\"afrexai-annual-report-1.0.0.zip\"",
        "redirectLocation": null,
        "bodySnippet": null
      },
      "scope": "source",
      "summary": "Source download looks usable.",
      "detail": "Yavira can redirect you to the upstream package for this source.",
      "primaryActionLabel": "Download for OpenClaw",
      "primaryActionHref": "/downloads/watcha-finder"
    },
    "validation": {
      "installChecklist": [
        "Use the Yavira download entry.",
        "Review SKILL.md after the package is downloaded.",
        "Confirm the extracted package contains the expected setup assets."
      ],
      "postInstallChecks": [
        "Confirm the extracted package includes the expected docs or setup files.",
        "Validate the skill or prompts are available in your target agent workspace.",
        "Capture any manual follow-up steps the agent could not complete."
      ]
    }
  },
  "links": {
    "detailUrl": "https://openagent3.xyz/skills/watcha-finder",
    "downloadUrl": "https://openagent3.xyz/downloads/watcha-finder",
    "agentUrl": "https://openagent3.xyz/skills/watcha-finder/agent",
    "manifestUrl": "https://openagent3.xyz/skills/watcha-finder/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/watcha-finder/agent.md"
  }
}
```
## Documentation

### Watcha AI Product Finder

You have access to the watcha.cn API — a Chinese AI product discovery platform with 1000+ products, user reviews, and community discussions. Your job is to help the user find AI products that genuinely fit their needs, not just the most popular ones.

### Core Principle: Popularity ≠ Quality

The watcha.cn community has biases you need to account for:

Review count and reply count reflect how talked about a product is (热度/hype), not how good it is. A niche but excellent tool may have 2 reviews; a mediocre but well-marketed tool may have 50.
Scores (stats.score) are only meaningful when review_count is substantial (roughly 10+). A score of 9.0 from 2 reviews tells you almost nothing. A score of 7.5 from 40 reviews is much more informative.
score_revealed being false means the score isn't shown publicly yet (too few reviews). Treat these products as "unproven" rather than "bad."
Upvotes vs downvotes can hint at community sentiment but are gameable.

Because of these limitations, always supplement watcha data with web searches to get a fuller picture — especially for products with few reviews.

### API Reference

All requests go to https://watcha.cn/api/v2/. Use these headers:

accept: application/json, text/plain, */*
content-type: application/json; charset=UTF-8
origin: https://watcha.cn
referer: https://watcha.cn/products
user-agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/144.0.0.0 Safari/537.36

### 1. Search Products

POST /search/general?q={query}&skip={offset}&limit={count}
Body: {"options":{"domains":["product"],"product_options":{"facets":["category_ids","tag_ids"]}}}

Filtering — add to product_options:

"category_ids": [6] — filter by category
"tag_ids": [4] — filter by tag

Search is exact-match, not fuzzy. If the user says "video editing AI", try multiple queries:

The exact product name if known
English keywords: "video editor", "video", "editing"
Chinese keywords: "视频编辑", "视频创作", "视频"
Or skip the query entirely (q=) and filter by category instead

When a text query returns few/no results, fall back to category browsing with q= (empty) and the relevant category_ids.

Categories:

IDNameEnglish1通用助手General Assistant2写作辅助Writing3图像生成Image Generation4视频创作Video Creation5音频处理Audio Processing6编程开发Coding/Dev7智能搜索Smart Search8知识管理Knowledge Management9科研辅助Research10智能硬件Smart Hardware11虚拟陪伴Virtual Companion12其他类型Other13Agent 构建Agent Building14效率工具Productivity153D 生成3D Generation

Tags (for tag_ids):

IDNameGroup2小程序 (Mini Program)平台形态3CLI平台形态4Web平台形态5移动端 (Mobile)平台形态6桌面端 (Desktop)平台形态8完全免费 (Free)商业费用9免费增值 (Freemium)商业费用10买断制 (One-time)商业费用12中国大陆 (China)可用地区13海外 (Overseas)可用地区

### 2. Product Detail

GET /products/{id_or_slug}

Returns full product info including description, organization, website_url, categories, stats, and tag.

### 3. Product Reviews

GET /products/{id}/reviews?order_by=score&replies=0&skip=0&limit=20

Reviews contain rich text in content.content (array of paragraphs → text nodes). Extract text by walking the structure. Each review has:

vote_value: 1 (upvote) or -1 (downvote) — the reviewer's sentiment
stats.upvotes: how many people found the review helpful
reply_count: discussion underneath
content.images: screenshot URLs (semicolon-separated)

### 4. Product Posts/Comments (Community Discussion)

GET /products/{id}/posts?order_by=newest&skip=0&limit=20

Posts are community discussions — feature requests, bug reports, invite code sharing, etc. They're useful for gauging community engagement but often contain noise (invite code begging, etc.). Skim them for substantive feedback, don't treat them as reviews.

### Workflow

When the user asks about AI products, follow this process:

### Step 1: Understand the need

Clarify what the user actually wants. Key dimensions:

Use case — what problem are they solving?
Platform — web, mobile, desktop, CLI?
Region — need China access? Or overseas only?
Budget — free, freemium, paid?
Specific features — e.g., "needs to support local models", "must have API"

### Step 2: Search broadly

Use the search API with multiple strategies to cast a wide net. The search is not fuzzy — be creative with queries:

Try the most specific keyword first
Try Chinese equivalents
Try broader terms
Fall back to category browsing if text search is unproductive

Fetch at least 10–20 results per search. Pagination: use skip and limit to page through results.

### Step 3: Shortlist candidates

From the search results, pick 3–5 candidates based on:

Relevance to the user's stated need (from the slogan and category)
Signal strength — products with more data points (reviews, upvotes) give you more to work with
Include at least one "dark horse" — a less-popular product that looks interesting based on its description

### Step 4: Deep-dive on shortlisted products

For each shortlisted product:

Fetch the product detail to read the full description
Fetch reviews (up to 20) — read the actual review text, not just the scores
Optionally fetch posts if you want community color
Search the web for the product name to get external perspectives — this is especially important for products with few watcha reviews. Check official websites, tech blogs, social media discussions.

### Step 5: Synthesize and recommend

Present your findings with nuance:

## [Product Name]
- **What it does**: one-line summary
- **Watcha score**: X.X (based on N reviews) — or "not enough reviews for a reliable score"
- **Community sentiment**: brief summary of what reviewers actually said
- **External info**: what you found from web searches
- **Best for**: who should use this
- **Watch out for**: any downsides or limitations mentioned

Rank by genuine fit for the user's needs, not by watcha score. Explain your reasoning.

### Step 6: Compare if asked

If the user wants to compare specific products, create a side-by-side table covering:

Core features
Pricing model
Platform availability
Community sentiment
Your assessment

### Tips

When the user asks a vague question like "推荐一些好的AI工具", ask a clarifying question about their use case before diving in.
For product names in Chinese, the slug field is often a romanized version you can use for web searches.
The website_url in product detail is the official site — useful for checking if the product is still active.
Review text is nested: content.content[].content[].text — walk the tree to extract it.
Some reviews are genuine and detailed; others are one-liners or invite code requests. Weight detailed reviews more heavily.
The hot_score field reflects trending momentum — useful for finding what's buzzing right now, but remember: hype ≠ quality.
## Trust
- Source: tencent
- Verification: Indexed source record
- Publisher: Charipoter
- Version: 0.1.0
## Source health
- Status: healthy
- Source download looks usable.
- Yavira can redirect you to the upstream package for this source.
- Health scope: source
- Reason: direct_download_ok
- Checked at: 2026-05-07T17:22:31.273Z
- Expires at: 2026-05-14T17:22:31.273Z
- Recommended action: Download for OpenClaw
## Links
- [Detail page](https://openagent3.xyz/skills/watcha-finder)
- [Send to Agent page](https://openagent3.xyz/skills/watcha-finder/agent)
- [JSON manifest](https://openagent3.xyz/skills/watcha-finder/agent.json)
- [Markdown brief](https://openagent3.xyz/skills/watcha-finder/agent.md)
- [Download page](https://openagent3.xyz/downloads/watcha-finder)