MCP Giga Checker
MCP server that checks text for signs of AI generation through the GigaChat API
Install
"mcpServers": {
"mcp-giga-checker": {
"command": "uvx",
"args": ["--from", "mcp_giga_checker", "mcp-giga-checker"],
"enabled": true,
"env": {
"GIGACHAT_AUTH": "ваши_авторизационные_данные_GigaChat",
"GIGACHAT_SCOPE": "GIGACHAT_API_CORP"
}
}
}Add to your MCP client's JSON config.
This is third-party code. Review the repository files before installing.
What it does
The server adds one scenario for the agent: send text to the GigaChat API and get an assessment of whether it looks AI generated. The check uses Sber's detection method rather than the agent model's own judgment. It is added as one MCP config entry via uvx and works in Cursor and other stdio clients.
Who it is for. Editors, teachers and HR staff at companies with corporate GigaChat API access.
Good fit when
- You need to check a submitted text for machine generation
- You want the check inside the agent workflow without a separate service
Not a fit when
- You are on a personal plan without corporate access
- The check result would be the only basis for a decision about a person
Example request
Check whether this student essay was generated by AILimitations
Generation detection is only available on corporate GigaChat API plans, and the config needs GIGACHAT_SCOPE set to GIGACHAT_API_CORP. Python 3.12 is required. The repository has not been updated since May 2025, and the PyPI package is published from the author's original repo trashchenkov/mcp_giga_checker. Any detector can be wrong, so treat results as a hint.
How to disable. Remove the mcp-giga-checker block from the MCP config or set enabled to false.
MCP
- Transport
- stdio
- Authentication
- API key
| Environment variables | |
|---|---|
| GIGACHAT_AUTH required, secret | GigaChat API authorization credentials |
| GIGACHAT_SCOPE required | Access scope; GIGACHAT_API_CORP for detection |
Security check
- Checked texts are sent to the GigaChat API
- Calls consume the corporate balance
README in short
The short Russian README explains the purpose and warns up front about plans: detection is for legal entities only. It lists requirements, Python 3.12 and GigaChat API access, and gives an MCP config snippet running via uvx with two environment variables. MIT license.
FAQ
Does it work with a personal GigaChat key?
No, according to the README detection is only available to legal entities on corporate plans.
Related
Microsoft's utility and MCP server that turn PDF, Word, Excel, PowerPoint, HTML and other files into Markdown for models
Anthropic skills
Skills
Anthropic's official skill examples: docx, pdf, pptx and xlsx documents, design, MCP builder and the Agent Skills spec
A skill that builds editable PPTX decks from PDF, DOCX, URLs and Markdown with native shapes, charts and animations
Anthropic's legal plugin suite: contracts, privacy, employment, IP, litigation, AI governance and legal education