An API for AI-generated code detection.
Send code, get back a per-file probability that it was written by AI, the regions that drove the score and an estimate of the model family. The same check runs peer and web plagiarism detection, so one integration covers both.
AI detection is included in every check at no extra credit.
Put AI detection where your users already submit code.
Learning platforms, assessment tools and hiring products collect code every day. The API adds AI-generated code detection to that flow without building or training a model of your own.
Per-file results
Every file gets an AI probability, a classification and a plain-language summary your UI can show as is.
Plagiarism in the same call
Peer similarity and web and GitHub matching run on the same check, with line ranges and linked sources.
SDK, CLI and MCP
An official Node.js SDK, a command-line client and an MCP server for AI assistants.
Structured results your product can render.
JSON with stable field names, per submission and per file, so you can show a score, a highlight or a full report.
AI detection fields
For each submission, an average and maximum AI probability and a risk level, and for each file its probability, classification and summary.
- ai_probability per file
- risk_level per submission
- Flagged regions and model family
Plagiarism fields
Peer and web scores per submission, matched files, similarity and line ranges for side-by-side views.
- Peer and web scores
- Matched line ranges
- Source URLs for web matches
Quick Check
One call that creates, uploads and starts a check, for simple integrations.
Status polling
Poll a check's progress and read results as soon as it completes.
Key-based auth
One API key in a header, with rate limit headers on every response.
From a ZIP to a JSON verdict.
Create a check
Name it, pick a language or auto detect, and turn AI detection on.
Upload submissions
One ZIP per submission, or many at once with batch upload.
Start the analysis
AI, peer and web checks run together on the engine.
Read the results
Fetch AI results and plagiarism results as JSON.
The whole flow in curl.
Replace the key and check id and this runs as is. Every call returns JSON; the last one returns the per-file AI detection results.
# 1. Create a check with AI detection on curl -X POST https://codequiry.com/api/v1/check/create \ -H "apikey: $CODEQUIRY_API_KEY" -H "Content-Type: application/json" \ -d '{"name":"Batch 12","language":999,"ai_run":true}' # 2. Upload one ZIP per submission curl -X POST https://codequiry.com/api/v1/check/upload \ -H "apikey: $CODEQUIRY_API_KEY" \ -F check_id=12345 -F [email protected] # 3. Start it curl -X POST https://codequiry.com/api/v1/check/start \ -H "apikey: $CODEQUIRY_API_KEY" -H "Content-Type: application/json" \ -d '{"check_id":12345,"ai_run":true}' # 4. Read the AI results when it finishes curl "https://codequiry.com/api/v1/ai-results?assignment_id=12345" \ -H "apikey: $CODEQUIRY_API_KEY"
Scores your users can act on.
A bare percentage invites the wrong decision. Each result carries its classification and a summary of what drove it, so your product can show evidence instead of a verdict.
The AI detection API, answered.
What does the AI detection API return?
For each submission, an average and maximum AI probability and a risk level; for each file, its AI probability, a classification such as "Strong AI-pattern match" and a short summary. Peer and web plagiarism results for the same check come from the results endpoints.
Does AI detection cost extra?
No. AI-generated code detection is included in every check at no extra credit. You pay for the check, not for each detector.
How accurate is it?
In held-out testing, model v3e caught 90% of AI-written code and falsely flagged 1.3% of human-written code. Show the score with its summary rather than as a pass or fail.
Which languages are supported?
More than 65 programming languages, with automatic language detection when you pass language 999 on check creation.
Are there SDKs?
Yes. There is an official Node.js SDK, a command-line client and an MCP server, and any language that can make an HTTP request can call the REST API directly.
Add AI detection to your product this week.
Get an API key, run the four calls above and render the results in your own interface.
More on AI-written code
Product names such as ChatGPT, Claude, GitHub Copilot, Cursor and Gemini are trademarks of their respective owners. Codequiry is an independent service and is not affiliated with, endorsed by, or sponsored by any of them.