Skip to content
MCP server

Rechtssysteem.ai

MCP server giving AI agents outcome predictions for Dutch court cases, with a leakage

Platforms
API
Version
0.3.2
Listing updated
today

Similar MCP servers

About Rechtssysteem.ai

Rechtssysteem.ai is an MCP server for legal and government. It is available on API.

Rather than returning a verdict on its own, this server exposes a model that estimates whether a Dutch court case will be rejected, partly granted or upheld, with probabilities per outcome. The README says the training data was cleaned to remove sentences that reveal the result before measuring accuracy, and it reports a majority-class baseline next to the model's score so the figures can be read in context.

The server runs over MCP, so an AI client can call the prediction tool, pull the published benchmark numbers, or test a text for outcome leakage. The leakage check and benchmark tools work without a key. The prediction tool uses a shared trial key with a rate limit of 20 requests per minute, and the README notes that case text is sent to the vendor's API for analysis.

It is aimed at developers and researchers who want to build legal analytics into an agent workflow or evaluate outcome-prediction datasets. The README states that the output is statistical and not legal advice, and that the tool answers 'undetermined' when confidence falls below 55%.

Tools this server provides (3)

Taken from the project documentation; the running server may differ.

voorspel_uitkomst
Predicts whether a Dutch case will be rejected, partly granted or granted, with probabilities per class.
rechtspraak_cijfers
Returns the published benchmark figures for the prediction model.
lekkage_check
Checks whether a text already reveals the court's outcome.

What it does

Outcome probability prediction
Predicts a Dutch case outcome as rejected, partly granted or granted, with probabilities for each class.
Leakage check
Measures whether a text already reveals the court's decision, usable on other datasets or AI claims.
Published benchmark figures
Returns accuracy, macro-F1 and majority baseline results from the model's validation.
Uncertainty threshold
Returns an undetermined answer when statistical confidence is below 55 percent.
Keyless local tools
Leakage check and benchmark tools run without an API key.
Single-file or package install
Can be run with uvx, pipx or as one Python file that uses only the standard library.

Good for

  • Adding a court outcome estimate to a legal research agent
  • Checking whether a legal dataset leaks its labels before training a model
  • Benchmarking Dutch case outcome predictors against a majority baseline
  • Testing AI-generated legal claims for hidden outcome information

Rechtssysteem.ai FAQ

Does the leakage check need an API key?

No. The README says lekkage_check and rechtspraak_cijfers work without a key, while voorspel_uitkomst requires one.

Where is case text processed?

The README says the case text is sent to the Rechtssysteem.ai server for analysis, with a maximum of 20,000 characters per request, so users should not send text they are not allowed to share.

Is the output legal advice?

No. The README states that the output is not legal advice, not a binding assessment of case chances, and not a replacement for a lawyer.

What happens when the model is unsure?

Below a 55 percent confidence level the tool returns an undetermined result, which the README describes as the intended behaviour.

What are the best alternatives to Rechtssysteem.ai?

The closest alternatives to Rechtssysteem.ai are LexITanon, Pharma, India Law, RegAI Legal MCP (Taiwan) and Pandex.

Where this comes from

  • Description: written from the project's README, the project's website and its MCP Registry entry.

How AIPoweredBot checks servers