SF Config Debt Radar
SahirVhora/sf-config-debt-radar
Detector de débito configuracional em SAP SuccessFactors para agentes de IA
Ver no GitHubScanner 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-radar1 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
- 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.
What It Does
- Pulls OData
$metadatawith 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
$countchecks 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
| Tool | What it does |
|---|---|
sf_scan_metadata_xml | Analyse raw $metadata XML for config debt (offline) |
sf_test_connection | Test OData v2 connectivity + get entity counts |
sf_scan_tenant | Full live tenant scan - pulls metadata + $count checks |
sf_assessment_questions | Generate guided config debt workshop questions |
sf_rate_findings | Score a set of findings with debt score + roadmap |
sf_known_ec_entities | List entity patterns the scanner recognises |
sf_about | Server 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