convez.ai
MCP ServerIntegração & APIsIntermediário
lociq-data

LOCIQ Agent Toolkit

lociq-data/agent-toolkit

Agents que entendem grafos de relacionamento de propriedades e portfolios imobiliários

Ver no GitHub
InicialPluginScore · de 0 a 100

MCP server e toolkit de referência para construir agents de IA que consultam dados de propriedades nos EUA (proprietários, negócios, parcelas e relacionamentos). Inclui prompt de sistema, implementação Python minimalista (~150 linhas) e cookbook de workflows prontos para pesquisa imobiliária, inteligência de mercado e análise de portfólios.

Instalação

$npx convez-ai lociq-data/agent-toolkit

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
0
Acompanhando
0
Saúde
50%
Último commit
há 4 meses
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Pesquisa de investimentos imobiliários e LLC clusters
  • Inteligência de mercado sobre portfólios de propriedades
  • Análise de relacionamentos entre proprietários e negócios ligados a imóveis
  • Automação de workflows de due diligence imobiliária

Veredito

Toolkit bem pensado que resolve um problema real: como agents raciocinam sobre grafos de dados imobiliários. A abordagem de system prompt + referência minimalista + cookbook é didática e prática. Status: ainda em construção (scaffolding em maio/2026), mas a direção é sólida para quem trabalha com dados imobiliários via LOCIQ.

README

A tradução para o português deste README ainda está na fila.

LOCIQ Agent Toolkit

Build AI agents that answer questions about U.S. property data — ownership, businesses, parcels, and the relationships between them.

This toolkit is for developers and AI engineers using LOCIQ as a data source for their agents. It ships:

  • A system prompt that teaches an agent how to think about LOCIQ data — confidence tiers, evidence model, the cluster graph, and how to compose tools into workflows
  • A Python reference implementation that connects an agent to LOCIQ's MCP server in roughly 150 lines, heavily annotated as a teaching artifact
  • A cookbook of complete, runnable workflows — investor research, market intelligence, evaluation harnesses — that your agent can read and execute

You bring your own LLM (Claude, GPT, Gemini, whatever). You bring your own agent framework or write the loop yourself. LOCIQ provides the data, the tools, and the workflows. Your agent does the rest.

Status (2026-05-13)

This repo is being built in the open. Today there's scaffolding and the documents you're reading. SYSTEM_PROMPT.md and the Python reference implementation are landing in coming days; cookbook entries follow. Star the repo to track. File an issue if you want a specific workflow in the cookbook.

What's in here

agent-toolkit/
├── prompts/           SYSTEM_PROMPT.md — the conceptual document teaching agents how to think about LOCIQ data
├── reference/         agent.py — minimum-viable Python agent connecting to LOCIQ's MCP server
└── cookbook/          Workflow recipes: problem statement + complete code + expected output + gotchas

Quickstart (coming soon)

The quickstart below describes the target shape. It does not run today.

git clone https://github.com/lociq-data/agent-toolkit.git
cd agent-toolkit
pip install -r reference/requirements.txt
export LOCIQ_API_KEY=your_key
export ANTHROPIC_API_KEY=your_key
python reference/agent.py "Find every property owned by Big Sky Properties LLC and its related entities"

The reference implementation is intentionally small. Read it, understand it, then build the agent you actually need.

Why this exists

LOCIQ's relationship graph — portfolios, LLC clusters, business chains, mixed-property owners — is the moat. The graph is queryable through standard API endpoints, but agents trip over the shape: which tool to call first, how to compose tools, what the confidence tiers mean, when "I don't know" is the right answer.

This toolkit is the answer to that. The system prompt teaches the shape. The reference implementation shows the integration. The cookbook ships proven workflows so your agent doesn't have to discover them.

Contributing

See CONTRIBUTING.md. Issues and pull requests welcome; LOCIQ does not accept custom-development requests — read CONTRIBUTING.md before filing a feature request.

License

MIT — see LICENSE.

Ver o original no GitHub

Bom para

Relacionados