xby-advertising-analysis
Run the following command in DeepSeek Harness:
dsh plugin install xby-skill/xby-advertising-analysis
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install xby-skill/xby-advertising-analysis inside DeepSeek Harness to install this plugin. Source repository: https://github.com/xby-skill/xby-advertising-analysis
About this plugin
As the LLM application ecosystem accelerates, the security boundaries at the middleware layer are easily overlooked. xby-advertising-analysis is an MCP server demo built for the DeepSeek Harness ecosystem, whose primary goal is not to inject advertisements but to fully expose the mechanics of ad-injection middleware so developers can observe, in a controlled environment, how ad content gets embedded into the LLM response pipeline.
The project offers persistent API key configuration and implements a chain of tool endpoints including get_completion, analyze_code, and developer_tip. Through these interfaces, developers can trace how data flows between requests and responses, identify at which stages a middleware can tamper with or inject content, and map out the potential attack surface.
It is best suited for security researchers, LLM application architects, and engineers curious about the MCP protocol and middleware architecture. If you are designing defenses against ad hijacking or content injection, this project provides a low-cost sandbox that shows you exactly what an attacker would do, so you can get ahead of it.
Use Cases
- Observe ad-injection middleware behavior in a controlled sandbox
- Study data flow and tampering risks under the MCP protocol
- Evaluate attack surfaces before designing anti-hijack defenses for LLM apps
Best For
- LLM application security researchers
- LLM architects focused on middleware architecture
- Engineers interested in MCP protocol and middleware security
Related Plugins
A network security red-team benchmark plugin for DeepSeek-V4.1/Flash featuring dual-layer kernel injection and zero-tool architecture for controlled model safety assessment.
A security red-team prompt plugin for DeepSeek-V4 that uses code-formal mapping and zero-trigger system prompts to enable unconditional output, intended for authorized testing and research only.
Adds an Auto approval mode on the official workspace-write sandbox, classifying semantic risks, asking when ambiguous, and denying destructive operations.
Undo/rollback system for DSH: config & plugin snapshots, one-click undo/redo/restore, message-level revert, secret masking, safe mode, and offline WebUI/GUI/CLI rescue tools for when DSH won't start.