O2C Deployment Workbench MCP Server
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@O2C Deployment Workbench MCP Server Can you investigate payment SAP-P001-09 and explain the exception?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
O2C Deployment Workbench
A complete, clearly synthetic customer implementation for ERP receivables: source onboarding, canonical data, cash application, deductions, collections, controls, KPI benchmarking, MCP access, multi-agent exception review, observability, UAT, cutover, and hypercare.
All SAP S/4HANA-style, Oracle Fusion-style, and NetSuite-style files in this repository are generated simulations. They are not genuine exports or live integrations. KPI movement is a controlled implementation benchmark on synthetic data, not customer impact.
Why this is an implementation project
The workbench starts with customer configuration and three incompatible source layouts, not a clean analytics table. It applies validation gates, preserves payment residuals, routes exceptions, prioritizes collections, shows unmet targets, and produces the artifacts a deployment team would use from discovery through go-live.
The architecture deliberately keeps financial matching deterministic and auditable. Agents handle judgment-heavy work: exception investigation, data-quality diagnosis, collection planning, and go-live monitoring. They cannot post to an ERP, issue a refund, write off debt, or alter master data.
flowchart LR
A["Three simulated ERP export sets"] --> B["Configuration-driven adapters"]
B --> C["Seven canonical receivables tables"]
C --> D["Validation control gate"]
D --> E["Deterministic cash matching"]
E --> F["Exception and collections queues"]
F --> G["MCP evidence tools"]
G --> H["Supervisor agent"]
H --> I["Cash Application Agent"]
H --> J["Data Quality Agent"]
H --> K["Collections Agent"]
I --> L["Human approval gate"]
J --> L
K --> L
E --> M["Controlled KPI benchmark"]
M --> N["Go Live Monitor Agent"]
O["OpenTelemetry traces"] -.-> B
O -.-> E
O -.-> H
O -.-> NRelated MCP server: syntheticcfo
Run it
python -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"
o2c-workbench --project-root .
streamlit run app.pyThe reproducible seed-42 run creates 1,908 canonical records, detects 18 intentional duplicate bank transactions as warnings, and leaves zero validation errors. Open the static dashboard or use the Streamlit app.
MCP server
The server uses the current MCP Python SDK and exposes structured tools, resources, and prompts. Local clients should use stdio; deployed clients can use Streamable HTTP.
o2c-mcp --transport stdio
o2c-mcp --transport streamable-http --port 8000Tools: get_implementation_summary, investigate_payment, get_collection_priority, and run_controlled_benchmark. Resources: o2c://configuration, o2c://benchmark, and o2c://observability.
Multi-agent paths
The default path is offline and deterministic so a reviewer can reproduce handoffs and controls with no account or API key. It produces output/agent_decisions.csv, output/agent_control_summary.json, and OpenTelemetry spans.
An optional model-backed review uses the OpenAI Agents SDK with specialist handoffs and structured output:
pip install -e ".[ai]"
export OPENAI_API_KEY="..."
o2c-agent-review SAP-P001-09 --project-root .The model path is intentionally not required for benchmark KPIs. It can recommend but cannot execute controlled financial actions.
Repository map
config/— implementation configuration, mappings, tolerances, weights, routes, controls, KPI targets, agent limits.data/raw/— generated, visibly labeled simulated ERP exports.data/canonical/— normalized customers, invoices, payments, remittances, deductions, promises, and activities.src/o2c_workbench/— adapters, matching, collections, controls, agents, MCP, observability, KPI logic.output/— benchmark evidence, queues, traces, and dashboards.docs/— process designs, mapping, controls, RTM, 28 UAT cases, cutover, monitoring, training, and research basis.deliverables/— implementation workbook.demo/— five-minute narrated implementation walkthrough and script.
Seed-42 benchmark
KPI | Current state | Configured future state | Movement |
Auto-match rate | 10.0% | 80.7% | +70.7 pp |
Manual-review rate | 90.3% | 21.9% | -68.5 pp |
Unapplied cash | $2,192,667 | $550,209 | -$1,642,459 |
DSO | 115.9 days | 51.6 days | -64.3 days |
CEI | 8.7% | 75.7% | +67.0 pp |
Past-due AR | 69.5% | 44.9% | -24.6 pp |
Processing time | 67.7 hours | 13.0 hours | -54.7 hours |
Four stretch targets remain unmet, so the monitor recommends a conditional go, not a victory lap. That is intentional implementation realism: proceed only with human approval controls and focused hypercare for unapplied cash and residual overpayments.
Verification
pytest -qSee research basis for the primary sources that informed the payment cases, MCP transport choice, agent handoffs, and telemetry conventions.
Available Tools
4 toolsget_collection_priorityA
Return the configured collection score evidence for one synthetic invoice.
Inputs: invoice_id (required). Returns: object
get_implementation_summaryA
Return controlled benchmark KPIs, data-quality status, disclaimers, and agent readiness.
Inputs: no named parameters. Returns: object
investigate_paymentC
Return matching evidence and exception routing for one synthetic payment.
Inputs: payment_id (required). Returns: object
run_controlled_benchmarkA
Regenerate synthetic source exports and controlled benchmark outputs; never connects to a genuine ERP.
Inputs: seed. Returns: object
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 4 tools
Each tool targets a clearly distinct function: summary/KPIs, payment investigation, collection priority, and benchmark regeneration. There is no meaningful overlap or ambiguity between them.
All names use lowercase snake_case and follow a verb_object pattern, but the verb set is mixed (get, investigate, run). This is still predictable and readable, with only minor stylistic variance.
Four tools is a tight, well-scoped surface for a synthetic O2C benchmark workbench. Each tool serves a distinct purpose without unnecessary bloat or missing core functionality.
The tool set covers the main workflows: viewing benchmark status, investigating a payment, checking collection priority, and regenerating outputs. A minor gap is the lack of explicit discovery tools for available synthetic payment or invoice identifiers, but this may be addressed through the summary.
Maintenance
Related MCP Connectors
Governed retail, FMCG, and CPG operational tools: runs, cases, approvals, audit-ready execution.
Map how a company works, and run B2B order to cash with AI agents that ask before they act.
Tesorio's order-to-cash tools — query customers, invoices, receivables, payments, and more.
Quote-to-cash tools: parse invoices and POs, match POs to invoices, collection emails, billing math.
Related MCP Servers
- FlicenseAqualityCmaintenanceProvides AI agents with deterministic, offline finance tools for commodity margin analysis, loan covenant compliance, invoice auditing, AP exception classification, and five-day close readiness.143 npm-

syntheticcfoofficial
AlicenseNot gradedqualityCmaintenanceEnables MCP clients to generate synthetic ERP datasets, download the generated packages, and read labelled fraud answer keys using plain language.MIT- AlicenseNot gradedqualityAmaintenanceEnables AI agents to query and reconcile electronic invoicing cases across multiple editors using six standardized MCP tools, with synthetic data and conformance checks.MIT
- AlicenseNot gradedqualityCmaintenanceEnables natural-language-driven purchase-to-pay workflows against mock procurement and payment APIs, including invoice matching, exception handling, human approvals, and idempotent payment scheduling with a tamper-evident audit log.MIT