Systemonomic — Work Domain Analysis & AI Task Suitability
LocalAbout Systemonomic — Work Domain Analysis & AI Task Suitability
Systemonomic provides work-domain analysis, AI task-suitability assessment, and organizational design for teams modeling work and planning structures. Its MCP servers cover workflows from generating domain models to assessing tasks and exporting organization reports.
Key features
- 01Create and edit work-domain nodes and links
- 02Generate a work-domain model from text
- 03Derive tasks and suggestions from models
- 04Assess task AI suitability in batches
- 05Propose and validate organizational changes
- 06Export project data and PDF reports
Use cases
- 01Map procurement or another department's work
- 02Derive tasks and identify automation candidates
- 03Validate a RACI matrix and plan restructuring
- 04Save organization snapshots and generate PDF reports
MCP support
It provides MCP servers for work-domain analysis, ATSS task assessment, and organizational design, exposing tools for projects, tasks, assessments, and organization design to AI agents.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
uvx systemonomic-mcpSYSTEMONOMIC_API_KEYSecretYour Systemonomic API key (generate at systemonomic.com/profile)Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMESystemonomic MCP Servers
MCP (Model Context Protocol) servers that expose Systemonomic's Work Domain Analysis, ATSS assessment, and organizational design capabilities to AI agents (Claude Desktop, Cursor, Claude Code, etc.).
Quick Start
1. Install
pip install systemonomic-mcp
2. Get an API Key
- Log in to Systemonomic
- Go to Profile → API Keys
- Click Generate API Key
- Copy the key (starts with
sk_sys_) — it's shown only once
3. Configure
Set the environment variable:
export SYSTEMONOMIC_API_KEY="sk_sys_your_key_here"
Optionally, point to a different API endpoint (defaults to production):
export SYSTEMONOMIC_API_URL="https://your-dev-backend.up.railway.app"
4. Add to Claude Desktop
Add to your claude_desktop_config.json:
{
"mcpServers": {
"systemonomic-wda": {
"command": "python",
"args": ["-m", "systemonomic_mcp.wda_server"],
"env": {
"SYSTEMONOMIC_API_KEY": "sk_sys_your_key_here"
}
},
"systemonomic-atss": {
"command": "python",
"args": ["-m", "systemonomic_mcp.atss_server"],
"env": {
"SYSTEMONOMIC_API_KEY": "sk_sys_your_key_here"
}
},
"systemonomic-org": {
"command": "python",
"args": ["-m", "systemonomic_mcp.org_server"],
"env": {
"SYSTEMONOMIC_API_KEY": "sk_sys_your_key_here"
}
}
}
}
5. Add to Cursor
In Cursor Settings → MCP Servers, add each server:
- Name:
systemonomic-wda - Command:
python -m systemonomic_mcp.wda_server - Environment:
SYSTEMONOMIC_API_KEY=sk_sys_...
Repeat for atss_server and org_server.
Available Servers
systemonomic-wda — Work Domain Analysis
| Tool | Description |
|---|---|
list_projects | List all projects |
get_project_state | Get complete project state |
create_project | Create a new project |
get_project_stats | Get project statistics |
list_wda_nodes | List WDA nodes |
create_wda_node | Create a node at a WDA level |
update_wda_node | Update a node's label/level/description |
delete_wda_node | Delete a node |
list_wda_links | List means-ends links |
create_wda_link | Link two nodes |
delete_wda_link | Remove a link |
generate_wda | AI-generate a full WDA from a text description |
export_project | Export project as JSON |
import_wda | Import nodes and links |
systemonomic-atss — Assessment & Tasks
| Tool | Description |
|---|---|
list_tasks | List project tasks |
create_task | Create a task |
generate_tasks_from_wda | Auto-generate tasks from WDA Objects |
derive_task_suggestions | AI-derived task suggestions |
list_suggestions | List pending suggestions |
accept_suggestions | Promote suggestions to tasks |
run_atss_batch | Run ATSS assessment on all tasks |
get_atss_results | Get stored assessment results |
persist_atss_results | Save assessment results |
list_atss_runs | List assessment run history |
systemonomic-org — Organizational Design
| Tool | Description |
|---|---|
get_org_design | Get current roles, org units, allocations |
persist_org_design | Save org design |
propose_restructure | AI-generated restructure proposal |
apply_proposal | Apply a restructure proposal |
validate_raci | Validate RACI matrix |
create_org_snapshot | Create version snapshot |
list_org_snapshots | List snapshots |
export_org_design_json | Export as JSON |
generate_pdf_report | Generate comprehensive PDF report |
get_report_status | Check report readiness |
Example Conversations
"Help me model our procurement process"
You: Generate a WDA for our university procurement department. They handle purchase requests, vendor management, contract negotiation, and compliance with government regulations.
Claude: Uses
create_project→generate_wda→ returns the generated hierarchy
"Assess which tasks can be automated"
You: For the procurement project, derive tasks from the WDA and run an automation assessment.
Claude: Uses
generate_tasks_from_wda→run_atss_batch→ summarizes automation candidates
"Generate the full report"
You: Create a PDF report for the procurement project.
Claude: Uses
generate_pdf_report→ saves the PDF
Development
# Run a server locally for testing
cd mcp
pip install -e .
SYSTEMONOMIC_API_KEY=sk_sys_... python -m systemonomic_mcp.wda_server
# Use the MCP inspector
SYSTEMONOMIC_API_KEY=sk_sys_... mcp dev src/systemonomic_mcp/wda_server.py
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.