← Back to Skills Marketplace
tttt-bjgs

小红书反检测处理

by tttt-bjgs · GitHub ↗ · v1.0.0 · MIT-0
cross-platform ⚠ suspicious
68
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install xhs-anti-detection
Description
Post-processes Xiaohongshu AI images by cleaning metadata, adding subtle noise and color shifts, protecting text, and re-encoding to reduce AI detection risk.
Usage Guidance
This package implements exactly what it says (techniques for hiding AI-origin signals in images), so it is coherent but ethically and operationally risky. Before installing or enabling: 1) Understand the legal/terms-of-service risk — these techniques are explicitly for evasion and may violate Xiaohongshu/other platforms' rules or local law. 2) Note the registry metadata lists no required binaries but the scripts require system tools (exiftool, tesseract) and Python packages — install and review those separately. 3) Review and audit the hooks/post_generate.py and any suggested integration snippets: they ask you to run persistent watchers and to modify other skills' code, which will cause automated, ongoing processing of files. Only enable auto-processing if you fully trust the workflow. 4) The scripts fake EXIF fields (camera model, timestamps) and write metadata in-place or to copies — if you need provenance, do not use this. 5) Run the tool in an isolated/test environment first (use disposable content/accounts) and inspect the outputs and logs. 6) If you plan to use it, review hard-coded paths in examples and confirm there are no network calls or hidden endpoints in any omitted files; consider running a local security review or sandboxed test to confirm no exfiltration. 7) If you are uncomfortable with deception or potential policy violations, do not install/activate this skill.
Capability Assessment
Purpose & Capability
The name/description (post-process Xiaohongshu AI images to reduce detection) aligns with the provided scripts and configuration: metadata cleaning, pixel-level noise, color shifts, text protection, re-encoding, and verification are all implemented. The presence of hooks to integrate with an image-generation skill is consistent with the stated purpose.
Instruction Scope
The SKILL.md and code instruct the agent/user to: (a) run a persistent folder watcher (hooks/post_generate.py) that will automatically process new images, and (b) optionally modify the image-generation skill to call this hook (code snippet provided). That is cross-skill modification / automatic processing beyond a single, manual tool invocation and increases the operational blast radius. The docs also instruct using burner accounts for testing and skipping verification in auto-mode. The run steps operate on local files only; I found no code that transmits images or metadata to external endpoints, but the auto-monitoring and suggested integration into other skills are scope-expanding behaviors to be aware of.
Install Mechanism
There is no formal install spec in the registry (instruction-only), and the package contains multiple Python scripts. Dependencies are managed via pip/OS packages (Pillow, numpy, opencv-python, pytesseract; system tools exiftool/tesseract). No remote downloads or archive extraction from arbitrary URLs are used. This is lower-risk than arbitrary network fetches, but note that system-level tools (exiftool, tesseract) must be installed separately and the README/USAGE assume those are present.
Credentials
The registry metadata declares no required binaries or environment variables, but the documentation and requirements clearly rely on external system tools (exiftool, tesseract) and optional pyexiv2. That discrepancy (metadata says 'none' while scripts require these tools) is an incoherence. The skill does not request credentials or secrets, which is proportionate, but several example scripts and batch helpers include hard-coded user paths (e.g., /Users/tianqu/.deskclaw/...) that may leak assumptions about environment or cause surprising behavior if blindly run.
Persistence & Privilege
always:false (good) and the skill does not claim or require persistent platform-level privileges. However, it ships a watcher (hooks/post_generate.py) and documentation that encourages running it as a background monitor or integrating calls into another skill to auto-process generated images. That gives the skill the ability to run continuously on a host and to act automatically on new content if the user enables those features. The instructions also recommend editing another skill to call the hook (modifies other skill code), which is a higher-privilege operation and worth caution.
How to Use
  1. Make sure OpenClaw is installed (local or Docker)
  2. Run the install command in chat: /install xhs-anti-detection
  3. After installation, invoke the skill by name or use /xhs-anti-detection
  4. Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
初始版本:小红书 AI 生成图片反检测处理,支持元数据清理、噪声添加、色彩偏移、文字保护、重新编码和验证
Metadata
Slug xhs-anti-detection
Version 1.0.0
License MIT-0
All-time Installs 0
Active Installs 0
Total Versions 1
Frequently Asked Questions

What is 小红书反检测处理?

Post-processes Xiaohongshu AI images by cleaning metadata, adding subtle noise and color shifts, protecting text, and re-encoding to reduce AI detection risk. It is an AI Agent Skill for Claude Code / OpenClaw, with 68 downloads so far.

How do I install 小红书反检测处理?

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

Is 小红书反检测处理 free?

Yes, 小红书反检测处理 is completely free, licensed under MIT-0. You can download, install and use it at no cost.

Which platforms does 小红书反检测处理 support?

小红书反检测处理 is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).

Who created 小红书反检测处理?

It is built and maintained by tttt-bjgs (@tttt-bjgs); the current version is v1.0.0.

💬 Comments