Choosing the Right STE Interface
Symmetry offers five ways to interact with the tax engine. This guide helps every person on your team (engineers, analysts, support, and product) find the right tool for what they're actually trying to do.
All five interfaces run against the same production STE: same tax tables, same calculation logic, same results. The difference is how you reach it and what your workflow requires.
Quick reference
| I need to… | Use… |
|---|---|
| Build or run payroll calculations in a production system | STE Hosted / On-Premise SDK |
| Ask STE a question in plain English, no code | STE MCP |
| Understand why a specific paycheck calculated the way it did | STL Chat |
| Show employees a plain-English breakdown of their withholding | STL API |
| Write integration code with STE-aware IDE suggestions | STE Copilot or STL Copilot |
The STE API is the core tax calculation engine. It accepts structured payroll inputs and returns precise federal, state, and local withholding results. Your engineers use it to power calculations inside your payroll platform.
Best for: Engineers building or maintaining a payroll integration, and any automated workflow that needs structured output.
Not designed for: Interactive questions, human-readable explanations, or exploration by non-developers.
The STE MCP connects the tax engine to AI assistants your team already uses — Claude, Cursor, and any other tool that supports the Model Context Protocol. Instead of assembling a JSON payload, you describe what you want. The MCP handles the request and returns a real, traceable STE result.
Best for: Tax and compliance teams verifying jurisdiction rules or rate changes. Support teams reproducing a client's calculation to diagnose a discrepancy. Implementation teams validating address-to-jurisdiction mappings. Anyone who needs a real answer from STE without writing code to get it.
Not designed for: Production pipelines or high-volume batch calculations. The MCP is built for interactive, human-driven use.
Example prompts:
Verifying a jurisdiction's tax rates:
For pay date 2026-05-15, what taxes apply to a Pittsburgh, Allegheny
County address? I need the Local Services Tax and EIT rate.
Reproducing a client's reported discrepancy:
A client is reporting that Ohio SUTA 2025 isn't capping at $9,000.
Here's the calculation detail from their report — please run it
through STE and walk me through where the cap should have applied.
Checking locals for new employee addresses:
Look up the tax jurisdictions for these three addresses [paste].
For each, list the local taxes a new employee would owe and note
which are mandatory versus elective.
Modeling a multi-state scenario:
Run a $3,000 regular wage in PA, NY, and CA — semi-monthly pay,
single filing status, 1 exemption. Show me federal, state, and
local totals side by side.
Symmetry Tax Logic (STL) Chat is a browser-based interface for walking through a specific paycheck result in plain English. Where the STE MCP suits open-ended exploration and hypotheticals, STL Chat is focused on a single workflow: take a real calculation, interrogate it conversationally, and confirm it's correct.
Best for: Support teams fielding "why did my check look like this?" questions. Engineers reviewing an STE log file while implementing a new tax. Anyone who needs to verify a real calculation without writing code or reading JSON.
Not designed for: Hypothetical modeling or broad compliance research — use the STE MCP for those.
The STL API accepts paycheck context and returns a plain-English explanation you can surface directly in your product — inside a pay stub detail view, an HR portal, or a self-service paycheck screen. Employees get a clear, specific answer to "why was this amount withheld?" without contacting support.
Best for: Product and engineering teams building employee-facing paycheck experiences. Reducing inbound support volume around withholding questions.
Not designed for: Internal analyst workflows — use STL Chat instead.
STE Copilot and STL Copilot: smarter integration code
The Copilots are context files that load Symmetry's documentation directly into your AI coding assistant (Cursor, Claude Code, or any compatible IDE). When your engineers have the STE or STL Copilot active, autocomplete suggestions are grounded in the actual STE and STL schema — correct parameter names, valid values, real integration patterns.
Best for: Engineers actively writing STE or STL integration code who want accurate, API-aware suggestions without switching to documentation.
Side-by-side comparison
| STE Hosted / SDK | STE MCP | STL Chat | STL API | Copilots | |
|---|---|---|---|---|---|
| What it does | Runs calculations | Answers questions | Explains a result | Embeds explanations | Improves code suggestions |
| How you use it | REST API calls | AI chat interface | Browser UI | REST API calls | IDE / code editor |
| Who it's for | Engineers | Analysts, support, PMs, payroll ops | Support, payroll ops, engineers | Engineers | Engineers |
| Requires code | Yes | No | No | Yes | In-editor only |
A note on data and trust
The STE MCP, STL Chat, and STL API are all read-only. They cannot modify your payroll runs, employee records, or tax configuration — they are lenses on the engine, not controls for it.
When Symmetry updates tax tables or rates, all five interfaces reflect those changes at the same time. There is no secondary data source to keep in sync.
