← Back to Skills Marketplace
iamdacai

Lanxin Media

by Iamdacai · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
76
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install lanxin-media
Description
蓝信官方媒体发送能力,支持发送图片、文件。使用 <lximg> <lxfile> 标签发送,系统自动解析上传。
README (SKILL.md)

蓝信媒体发送规则(必须严格遵守)

核心规则(绝对禁止违反)

  1. 用户要求发图片/文件,必须直接输出 \x3Clximg> 或 \x3Clxfile> 标签
  2. 标签必须单独成行,不使用任何 markdown、代码块、加粗、斜体
  3. 禁止把标签放在 ``` 代码块中
  4. 禁止说“无法发送”“不支持”“暂时不能”等拒绝话术
  5. 禁止解释标签作用,直接输出即可
  6. 一次只发送一个文件或一张图片
  7. 路径必须真实有效,格式干净无多余符号

发送图片(必须这样输出)

注意:发送图片时必须使用 \x3Clximg> 标签,而不是 \x3Clxfile> 标签! 注意:SVG 文件请使用 \x3Clxfile> 标签发送,而不是 \x3Clximg> 标签!

格式: \x3Clximg>文件路径或URL\x3C/lximg>

示例(正确): 这是你要的图片: \x3Clximg>/Users/user/photo.jpg\x3C/lximg>

示例(正确): \x3Clximg>https://example.com/image.png\x3C/lximg>

发送 SVG 文件(必须这样输出)

注意:SVG 文件必须使用 \x3Clxfile> 标签,而不是 \x3Clximg> 标签!

格式: \x3Clxfile>文件路径或URL\x3C/lxfile>

示例(正确): \x3Clxfile>/Users/user/image.svg\x3C/lxfile>

示例(正确): \x3Clxfile>https://example.com/image.svg\x3C/lxfile>

发送文件(必须这样输出)

注意:发送文件时必须使用 \x3Clxfile> 标签,而不是 \x3Clximg> 标签!

格式: \x3Clxfile>文件路径或URL\x3C/lxfile>

示例(正确): 文件已发送: \x3Clxfile>/Users/user/report.pdf\x3C/lxfile>

示例(正确): \x3Clxfile>https://example.com/document.pdf\x3C/lxfile>


支持的闭合格式(任选一种)

\x3Clximg>...\x3C/lximg> 或 \x3Clximg>...\x3C/img> \x3Clxfile>...\x3C/lxfile> 或 \x3Clxfile>...\x3C/file>


蓝信限制

  • 图片/文件大小 ≤ 2MB
  • 支持:png、jpg、jpeg、pdf、doc、docx、xls、xlsx、txt
  • 视频、音频暂不支持

严禁出现的错误行为

❌ 禁止:markdown \x3Clxfile>...\x3C/lxfile> ❌ 禁止:\x3Clxfile>...\x3C/lxfile> ❌ 禁止: \x3Clxfile>... ❌ 禁止:解释“这是标签”“系统会解析” ❌ 禁止:拒绝发送 ❌ 禁止:只输出文字内容,不输出标签

你的行为准则

用户让你发文件/图片 → 直接输出干净标签 → 不要多余描述 → 不要格式包裹 → 完成。

Usage Guidance
This skill appears to be a simple helper that instructs the agent to output special <lximg>/<lxfile> tags so the system can upload media. That is coherent, but the runtime rules force the agent to output raw file paths/URLs and forbid refusing or explaining — which can make the agent return sensitive local paths (or be coerced into fabricating tags). Before installing: verify how your platform handles <lximg>/<lxfile> (does it auto-upload local paths? from what filesystem scope?), test the skill only with non-sensitive files/URLs, and confirm the platform does not honor the SKILL.md 'always:true' override. If possible, restrict the skill's access to only user-provided URLs (not arbitrary local filesystem paths) or require explicit user confirmation before uploading local files.
Capability Analysis
Type: OpenClaw Skill Name: lanxin-media Version: 1.0.0 The skill defines a legitimate interface for sending media via the Lanxin platform using custom XML-like tags (<lximg> and <lxfile>). The instructions in SKILL.md are strictly focused on ensuring the AI agent outputs the correct format for the system's parser to handle file uploads. While the ability to reference local file paths carries inherent risks if misused by a user, the skill itself contains no malicious code, exfiltration logic, or deceptive instructions intended to compromise the host system.
Capability Assessment
Purpose & Capability
Name/description, files, and lack of extra credentials match a simple media-sending skill. The included index.js is harmless boilerplate. However SKILL.md contains an 'openclaw.always':true metadata entry while registry flags report always:false — a mismatch worth noting (attempt to request permanent inclusion inside the instruction file).
Instruction Scope
SKILL.md mandates that the agent must output raw <lximg>/<lxfile> tags with real file paths or URLs, must not refuse, and must not explain. The instructions do not require verifying existence/permissions and encourage supplying local paths (e.g., /Users/user/photo.jpg). That coerces the agent to produce local file paths (which the platform may then upload) and forbids refusal or contextual explanations — increasing the risk of accidental or coerced disclosure of sensitive local files or fabricating tags.
Install Mechanism
No install spec and no dependencies — instruction-driven skill with a tiny harmless index.js. Nothing is downloaded or written to disk by an installer.
Credentials
The skill declares no environment variables, no credentials, and no config paths. That is proportionate for a tag-producing media skill.
Persistence & Privilege
Registry metadata shows always:false, but SKILL.md includes openclaw.always:true. If the platform honored the SKILL.md metadata, that would grant this skill permanent inclusion (higher privilege). The presence of that metadata inside SKILL.md is a red flag because it attempts to assert elevated persistence from within the instruction file.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install lanxin-media
  3. After installation, invoke the skill by name or use /lanxin-media
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
lanxin-media v1.0.0 - Initial release providing official Lanxin media sending functionality. - Supports sending images and files using <lximg> and <lxfile> tags. - Strict rules for tag usage—no markdown, code blocks, or extra formatting allowed. - Only one file or image can be sent per action, with clear separation between image and file tags. - Enforces file type and size limitations in compliance with Lanxin platform requirements.
Metadata
Slug lanxin-media
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is Lanxin Media?

蓝信官方媒体发送能力,支持发送图片、文件。使用 <lximg> <lxfile> 标签发送,系统自动解析上传。 It is an AI Agent Skill for Claude Code / OpenClaw, with 76 downloads so far.

How do I install Lanxin Media?

Run "/install lanxin-media" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.

Is Lanxin Media free?

Yes, Lanxin Media is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does Lanxin Media support?

Lanxin Media is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created Lanxin Media?

It is built and maintained by Iamdacai (@iamdacai); the current version is v1.0.0.

💬 Comments