ProSuite MCP Server
ProSuite/prosuite-mcp
Verificação de qualidade de dados geoespaciais via IA
Ver no GitHubServidor MCP que expõe a verificação de qualidade ProSuite para assistentes de IA como Claude. Permite executar validações de dados geoespaciais, carregar especificações QA em XML e rodar verificações ad-hoc contra workspaces com suporte a filtros espaciais e múltiplas condições.
Instalação
npx convez-ai ProSuite/prosuite-mcp1 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
Casos de uso
- Validação de qualidade em geodatabases e datasets SDE
- Detecção de erros de conectividade e integridade espacial
- Automação de testes de conformidade QA em pipelines de dados
Veredito

Ferramenta especializada para equipes de geomática e gestão de dados espaciais. Bem integrada ao ecossistema Claude, mas requer infraestrutura ProSuite local já operacional — não é plug-and-play.
README
A tradução para o português deste README ainda está na fila.
prosuite-mcp
MCP server that exposes Dira ProSuite quality verification to AI assistants (Claude, etc.).
Prerequisites
A running ProSuite Quality Verification Server reachable from the host where this server runs.
Configuration
| Environment variable | Default | Description |
|---|---|---|
PROSUITE_HOST | localhost | ProSuite service host |
PROSUITE_PORT | 5151 | ProSuite service port |
PROSUITE_SSL_CERT_PATH | — | Path to PEM certificate for TLS |
Usage
Windows users: see docs/windows-setup.md for a step-by-step guide including uv and Claude Code installation.
Both options below assume you create a project directory first:
mkdir mytest
cd mytest
uv init --python 3.12
uv add prosuite-mcp
Claude Code CLI
Register the server from inside mytest, then start Claude:
claude mcp add prosuite \
-e PROSUITE_HOST=localhost \
-e PROSUITE_PORT=5151 \
-- uv run prosuite-mcp
claude
The -- uv run prosuite-mcp tells Claude Code to start the MCP server via uv run in the current project, so prosuite-mcp is resolved from the local .venv. Run claude from the same mytest directory each time.
Copilot CLI
Register the server from inside mytest, then start Copilot:
copilot mcp add prosuite \
-e PROSUITE_HOST=localhost \
-e PROSUITE_PORT=5151 \
-- uv run prosuite-mcp
Tools
load_spec <path> — Loads a .qa.xml spec file. Call this at the start of a session with the path to your spec (local drive, OneDrive, network share). Replaces any previously loaded spec.
search_spec <query> [max_results] — Searches the loaded .qa.xml spec for conditions matching a natural-language query (English, German, French, Italian). Returns up to max_results (default 20) matching conditions with pre-filled condition_request blocks ready to pass directly to run_verification, plus the required_datasets list. Requires a spec to be loaded first via load_spec.
list_conditions [search] — Lists available quality conditions. Pass a keyword to filter by name or description.
describe_condition <name> — Shows the full docstring and parameter list for a condition, including which parameters expect dataset names vs. primitive values.
run_verification — Runs an ad-hoc quality verification against a workspace. Key parameters:
| Parameter | Type | Description |
|---|---|---|
model_catalog_path | string | Workspace path on the server (C:/data/my.gdb, .sde file, …) |
model_name | string | Logical name for the data model |
datasets | list | Feature classes/tables: {name, filter_expression?} |
conditions | list | Conditions to run: {condition, params} |
output_dir | string? | Server-side directory for Issues.gdb and HTML report |
envelope | object? | Spatial filter {x_min, y_min, x_max, y_max} |
Returns a summary with status, total_errors, and a per-condition breakdown.
Example
Once connected, you talk to Claude in plain language:
Check road connectivity in
C:/data/tlm.sde.
With a spec loaded, Claude calls search_spec to find the relevant pre-configured conditions from the .qa.xml file, then calls run_verification with the pre-filled parameters and returns a summary of errors per condition.
Without a spec, Claude uses list_conditions and describe_condition to find and configure conditions from scratch.
Development
uv sync --dev
uv run pytest
uv run ruff check src
uv run pyright src
License
MIT — see LICENSE.