mcpField
R
ragjur

ragjur

Remote
ragjur.aiLegal & Government0Updated 1w ago

About ragjur

ragjur connects legal professionals and researchers to Brazilian court decisions and jurimetrics through Claude and other MCP clients. Its tools cover search, statistical analysis, and citation verification; conversational answers are produced by the host model.

Key features

  1. 01Full-text search across court decisions
  2. 02Semantic search for similar decisions
  3. 03Topic and panel trend analysis
  4. 04Litigation and outcome statistics
  5. 05Judge and rapporteur profiles
  6. 06Citation verification

Use cases

  1. 01Find recent decisions by court and topic
  2. 02Compare outcomes across court panels
  3. 03Estimate litigation duration or outcomes
  4. 04Check citations in a legal document

MCP support

Provides a remote MCP server over Streamable HTTP at `https://mcp.ragjur.ai/mcp`. It exposes legal search and jurimetrics tools to MCP clients, with the available tool set dependent on the subscription plan.

Configuration

Remote servers need no local install. Pick your client, then copy the config or install in one click.

Endpointhttps://mcp.ragjur.ai/mcp
Copy install command
Claude CodeTerminal
Cursor.cursor/mcp.json
VS Code.vscode/mcp.json
Claude Desktopclaude_desktop_config.json
Windsurf~/.codeium/windsurf/mcp_config.json

Run commands in a terminal; paste JSON into the client’s MCP config and restart. Servers that need sign-in will prompt on first connect.

All endpoints
Streamable HTTPhttps://mcp.ragjur.ai/mcpAuth required

Tools

Details

From the project README

RAGJur — Claude Plugin

Brazilian legal RAG for Claude — search 112M+ court decisions from 90+ Brazilian tribunals.

💳 Paid subscription required (from R$ 197/month — see ragjur.ai/precos). Installation and authorization are free; using the tools requires a RAGJur subscription: ragjur.ai.

This repository contains the Claude plugin manifest for RAGJur. The plugin connects Claude to the RAGJur MCP server (remote, streamable HTTP) at https://mcp.ragjur.ai/mcp.

Install

In Claude Code:

/plugin marketplace add btechbrasil/ragjur-mcp
/plugin install ragjur@btechbrasil

No API key to paste. On the first tool call, your host (Claude, ZCode/GLM, or any MCP client with OAuth support) receives the server's authorization challenge and opens your browser for login (BeansTech ID). Authorize once and the connection reconnects automatically; your plan travels in the token and the tool list adapts (see Plans below).

What you get — 12 grounding tools

ToolWhat it does
buscar_jurisprudenciaFull-text search over 112M+ decisions (BM25, filters by tribunal/class/date)
similaridadeSemantic search — decisions similar to a case or ementa (embeddings)
panorama_temaTopic overview: outcome rate, result categories, trend
divergencia_turmasDivergence between panels/chambers on a topic
evolucao_jurisprudencialHow a thesis evolved over time (quarterly)
predicao_resultadoStatistical outcome prediction with confidence interval
tempo_tramitacaoLitigation-time statistics (median, percentiles)
perfil_relatorJudge/rapporteur profile on a topic
dna_turmaStatistically characteristic terms of a panel
clusters_tematicosCluster a topic into sub-issues
estrategia_juizFull strategic analysis of a judge for a topic
verificar_citacaoAnti-hallucination citation verification

📊 Plans: Essential (R$ 197/mo, 100 queries) unlocks the core research tools (buscar_jurisprudencia, similaridade, verificar_citacao, panorama_tema); Pro (500 queries) unlocks all 12. The server filters the tool list to your plan automatically.

🧠 The prose is yours: conversational Q&A and document drafting are done by the LLM of the platform you're in (Claude, GLM/ZCode, ChatGPT, Cursor), grounded by these retrieval and jurimetrics tools. The plugin intentionally ships no chat/draft tool — the synthesis layer is the host model, so answers come from the model you already pay for. (chat_juridico and gerar_peca remain available on the REST API for direct integrations.)

Coverage

STF, STJ, TST, TSE, TCU, all TRFs (1–6), all state courts (TJs), labor courts (TRTs), electoral courts, state audit courts (TCEs) and administrative tax appeal tribunals (CARF/DRJs/TARFs) — 112 million+ decisions and growing daily.

Example prompts
  • "Busque os acórdãos mais recentes do STJ sobre busca e apreensão em alienação fiduciária"
  • "Qual a taxa de provimento de recursos sobre dano moral no TJSP? Há divergência entre as câmaras?"
  • "Verifique se as citações desta peça existem"
  • "Redija uma contestação sobre despejo por falta de pagamento, lado réu, TJMG"

How it works

The plugin registers one remote MCP server (HTTP transport). Your queries are sent to mcp.ragjur.ai, which searches the Elasticsearch corpus and returns grounded results. The server is stateless — no conversation or query data is persisted after the session ends.

⏱️ Response times: search and analytics tools answer in seconds; similaridade and estrategia_juiz can take up to ~30–60 s on hard queries. The plugin declares a 110 s tool timeout (aligned with the server's own cap) so hosts whose default tool timeout is 30 s don't abort these calls mid-flight.

Privacy & data

  • Stateless MCP server (30-minute session TTL, nothing stored)
  • Data processed in Brazil (Alibaba Cloud São Paulo region)
  • LGPD-compliant: ragjur.ai/lgpd
  • Privacy policy: ragjur.ai/privacidade

Support

License

MIT (this repository — plugin manifest and documentation). The RAGJur service is a commercial product of Beans Tech.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.