convez.ai
MCP ServerDevOps & InfraestruturaAvançado
SahirVhora

SF Config Debt Radar

SahirVhora/sf-config-debt-radar

Detector de débito configuracional em SAP SuccessFactors para agentes de IA

Ver no GitHub
InicialPluginScore · de 0 a 100

Scanner de SAP SuccessFactors que detecta e avalia débito técnico em configurações através de análise de metadados OData. Funciona como CLI, dashboard web e MCP server para integração com agentes de IA, gerando relatórios de risco e roadmaps de remediação.

Instalação

$npx convez-ai SahirVhora/sf-config-debt-radar

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
42%
Último commit
há 5 dias
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Auditar complexidade de schemas em ambientes SAP SuccessFactors
  • Identificar campos customizados e entidades não documentadas
  • Gerar roadmaps de limpeza técnica e governança com scores de risco

Veredito

Ferramenta especializada e bem construída para times SAP SuccessFactors que precisam governar configurações em escala. MCP server integra-se naturalmente com Claude e agentes, trazendo análise técnica profunda a contextos de IA. Nicho específico, mas inestimável para esse segmento.

README

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

SF Config Debt Radar

SAP SuccessFactors EC configuration debt scanner. Analyse metadata, detect config debt, score risk - from your terminal, browser, or any MCP-compatible AI agent.

mcp python

What It Does

  • Pulls OData $metadata with Basic Auth or OAuth2
  • Parses EC, Position, Foundation Object, Picklist, and custom MDF entities
  • Scores hidden configuration debt from schema complexity, field counts, and null rates
  • Runs safe Tier 1 $count checks for null/blank signals
  • Produces JSON reports with debt score + 90-day remediation roadmap
  • Exposes everything as MCP tools for AI agent integration

Privacy

Zero employee data is stored or leaves the tenant. Schema metadata, entity counts, and non-identifiable signals only.


Quick Start

pip install -r requirements.txt

# Offline metadata demo (no tenant needed)
python -m sf_config_debt_radar metadata-demo --output output

# Live tenant scan
cp config.example.yaml config.yaml
# edit config.yaml with your connection
python -m sf_config_debt_radar scan --config config.yaml --output output

Browser Dashboard

python -m http.server 8088
# open http://localhost:8088/index.html

MCP Server (AI Agent Integration)

The MCP server lets any MCP-compatible AI agent (Hermes Agent, Claude Code, Cursor, VS Code) discover and call the scanner's tools automatically.

One-liner run

python mcp_server.py

Connect from any MCP client

Hermes Agent - add to ~/.hermes/config.yaml:

mcp_servers:
  sf-config-debt-scanner:
    command: /path/to/sf-config-debt-radar/run_mcp_server.sh
    enabled: true

Claude Code - add to ~/.claude/claude_desktop_config.json (or project .claude/settings.json):

{
  "mcpServers": {
    "sf-config-debt-scanner": {
      "command": "python",
      "args": ["/path/to/sf-config-debt-radar/mcp_server.py"]
    }
  }
}

Cursor - add under Settings > MCP Servers:

name: sf-config-debt-scanner
type: command
command: python /path/to/sf-config-debt-radar/mcp_server.py

Available MCP Tools

ToolWhat it does
sf_scan_metadata_xmlAnalyse raw $metadata XML for config debt (offline)
sf_test_connectionTest OData v2 connectivity + get entity counts
sf_scan_tenantFull live tenant scan - pulls metadata + $count checks
sf_assessment_questionsGenerate guided config debt workshop questions
sf_rate_findingsScore a set of findings with debt score + roadmap
sf_known_ec_entitiesList entity patterns the scanner recognises
sf_aboutServer metadata and version

Example prompts

"Scan this SF metadata XML for custom field sprawl" "Test connection to api55.sapsf.eu" "Run a full config debt scan against my tenant" "Give me assessment questions for governance"


Browser App

The index.html dashboard is best for:

  • Client workshops
  • Self-assessment sessions
  • Demo metadata import / paste
  • Team walkthroughs

Note: Live tenant calls from the browser usually need a CORS proxy. The Python scanner is the reliable live-tenant path.


Config Examples

Basic Auth:

sf:
  auth_method: basic
  base_url: "https://api55.sapsf.eu/odata/v2"
  username: "admin@COMPANYID"
  password: "..."

OAuth2:

sf:
  auth_method: oauth2
  base_url: "https://api55.sapsf.eu/odata/v2"
  client_id: "..."
  client_secret: "..."
  company_id: "COMPANYID"
  token_url: "https://api55.sapsf.eu/oauth/token"

Project Structure

sf-config-debt-radar/
  mcp_server.py              # MCP server (7 tools, stdio or SSE)
  run_mcp_server.sh           # Hermes launcher script
  index.html                  # Browser dashboard
  config.example.yaml         # Config template
  pyproject.toml              # Package definition
  requirements.txt            # Dependencies
  sf_config_debt_radar/       # Core library
    cli.py                    # CLI entry point
    auth.py                   # SF OData authentication
    metadata.py               # XML metadata parser
    scanner.py                # Scan engine
    scoring.py                # Debt scoring
    report.py                 # Report generation
    __main__.py               # python -m support
  tests/
    test_core.py
  data/                       # Client evidence templates
Ver o original no GitHub

Bom para

Relacionados