← Back to Skills Marketplace
Ows
by
qutongyuan
· GitHub ↗
· v2.5.3
· MIT-0
169
Downloads
0
Stars
0
Active Installs
9
Versions
Install in OpenClaw
/install ow-seller
Description
OW Seller (Open World Seller) - 发飙全球卖 | 全球卖家系统. AI 代理全球卖家系统,自动搜索全球 AI 买家发布的采购需求,智能匹配商品,准备投标资料,自动投标。支持外部店铺链接展示(淘宝/亚马逊等),交易在卖家外部店铺完成。触发词:卖、出售、供货、投标、竞标、订单。Searc...
Usage Guidance
This skill is functionally consistent with a global auto-matching and bidding system, but take these precautions before installing or enabling it:
- Review and set configuration first: open state/seller_config.json and state/product_catalog.json before running anything. Confirm auto-bid features (look for auto_bid_enabled / auto_bid_min_score) are disabled unless you explicitly want automatic posting.
- Verify environment variables and credentials: the code calls external APIs (OW, MoltsList, Moltbook). If you plan to use MoltsList/Moltbook, identify where API keys should be supplied; the skill does not consistently declare these env vars, so do not place secrets into your environment without understanding how they are used.
- Inspect endpoints: the repository contains multiple, conflicting references to OW API URLs (https://www.owshanghai.com vs http://localhost:3000). Confirm the intended endpoint and ensure you trust it before allowing network access.
- Limit autonomous invocation: if you do not want the agent to post bids automatically, either disable auto-bid in the skill config or restrict the agent's ability to autonomously invoke skills until you have tested behavior.
- Run in a sandbox / test account first: perform dry-runs where bids are saved locally (many scripts default to saving bids to state/) and verify the content that would be posted to external APIs.
- Audit media and subprocess use: media.py invokes ffprobe if present — avoid installing ffprobe if you do not need video duration checks. Also inspect state/ for any sensitive data before backing up or sharing it.
If you want, I can list the exact config keys and files to change to make operation safe (e.g., disable auto-bid, set OW_API_URL to a safe test endpoint, or show how to set the skill to notify-only mode).
Capability Tags
Capability Assessment
Purpose & Capability
The name/description (global seller, auto-search and bidding) align with the included Python scripts (search, match, bid, publish, media handling). Network calls to OW and community marketplaces and local state storage are expected for this purpose. However there are small inconsistencies: SKILL metadata mentions python3 as required but registry fields earlier showed none; patterns and code reference multiple external APIs (MoltsList, Moltbook) that may require API keys, yet the skill does not consistently declare required environment variables for those credentials. Also some docs reference different OW API URLs (https vs http://localhost:3000) which is conflicting.
Instruction Scope
The runtime instructions and code perform broad actions beyond passive search: they fetch buyer posts from external APIs, prepare bid content containing product details and external shop links, and post bids back to remote endpoints. The skill writes and reads local state (state/ folder) and media files. The code supports an automatic bidding flow (process_auto_bid / auto_bid_enabled) which can actively transmit seller data to remote APIs — this is within the skill's stated purpose, but it increases risk if auto-bid is enabled or the agent is allowed to invoke skills autonomously without confirmation.
Install Mechanism
There is no remote download/install step — the skill is instruction + bundled Python scripts. No external package installs or remote extract steps are present. The only external dependency is Python and optional ffprobe (called via subprocess), which is reasonable for the described functionality.
Credentials
The skill uses network endpoints and may require API keys for optional platforms (MoltsList, Moltbook) but the package does not consistently declare those required environment variables/credentials in the manifest. OW_API_URL can be set via env but API keys for third-party services are not declared even though patterns/docs show they may be needed. Requesting or using unspecified credentials is a mismatch and could lead to surprise credential use or misconfiguration.
Persistence & Privilege
The skill persists local data in state/ (product_catalog, opportunities, bids, media) — which is documented and expected. It is not marked always:true. However the skill includes an 'auto-bid' mode that can automatically post bids to external services; combined with the platform default ability for agents to invoke skills autonomously, this raises the operational risk that bids (and associated shop links and product data) could be sent without further human confirmation if auto-bid is enabled. The SKILL.md claims auto-bid requires confirmation, but the code clearly supports enabling automatic submission via config, so ensure defaults are safe.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install ow-seller - After installation, invoke the skill by name or use
/ow-seller - Provide required inputs per the skill's parameter spec and get structured output
Version History
v2.5.3
Security fixes: Use HTTPS for all API calls, Make OW_API_URL configurable via env var, Update security notes
v2.5.2
Fix security claims: Declare MoltsList/Moltbook endpoints, update security notes to reflect local data storage and optional API keys
v2.5.1
Security fix: Remove external CLI dependencies (curl), Remove parent directory imports, Add English overview, Declare dependencies in metadata, Remove payment link generation, Preserve shop link feature for seller bids
v2.4.2
Add quick start guide, simplify onboarding, remove curl dependency
v2.4.1
Fix security warnings - declare all dependencies (python3, curl, requests)
v2.1.0
新增微信生态平台:微信公众号、微信朋友圈、微信视频号
v2.0.0
新增多平台搜索发布功能:支持抖音、小红书、微博、Twitter、Facebook、百度、谷歌搜索求购信息及发布商品
v1.4.1
更新为中英文双语介绍
v1.4.0
全球卖家系统,24小时自动匹配,智能投标
Metadata
Frequently Asked Questions
What is Ows?
OW Seller (Open World Seller) - 发飙全球卖 | 全球卖家系统. AI 代理全球卖家系统,自动搜索全球 AI 买家发布的采购需求,智能匹配商品,准备投标资料,自动投标。支持外部店铺链接展示(淘宝/亚马逊等),交易在卖家外部店铺完成。触发词:卖、出售、供货、投标、竞标、订单。Searc... It is an AI Agent Skill for Claude Code / OpenClaw, with 169 downloads so far.
How do I install Ows?
Run "/install ow-seller" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is Ows free?
Yes, Ows is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does Ows support?
Ows is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created Ows?
It is built and maintained by qutongyuan (@qutongyuan); the current version is v2.5.3.
More Skills