How it works
A fixed lexicographic grid scan (last variable fastest) evaluates the claim expression <= bound or >= bound with an explicit tolerance, capped by an evaluation budget, and returns the first witnesses with their margins.
Inputs and units
Provide the expression, the relation (le or ge), the bound, optional tolerance, one to six variable domains (min, max, 2–64 steps) and optional evaluation and counterexample caps.
What the result contains
The scanned fraction, completeness, explicit stopping reason and numerical witnesses with values and margins. Undefined and near-boundary evaluations are counted separately and can leave the result inconclusive.
Example MCP call
{
"name": "scan_counterexample",
"arguments": {
"expression": "x ^ 2",
"relation": "le",
"bound": 4,
"variables": [
{
"name": "x",
"min": -3,
"max": 3,
"steps": 7
}
]
}
}Send this tool name and arguments through a connected MCP client. Discover the authoritative input schema with tools/list.
Call this engine in one command
Node.js 20+ and npm. This downloads a small remote MCP client; it does not install the engine locally.
npx --yes --package https://scorecompute.com/downloads/scorecompute-call-0.14.1.tgz scorecompute-call scan_counterexample --arguments '{"expression":"x ^ 2","relation":"le","bound":4,"variables":[{"name":"x","min":-3,"max":3,"steps":7}]}'Client download, PowerShell input and verification →
Direct MCP access
scan_counterexample is independently callable. Its generated registration deliberately disables orchestration: no planner binding or approved adapter is supplied. Do not infer composability from matching JSON fields.
Cite this engine
Version 0.2.0, source SHA-256 2df9bab83183…, release 0.16.1. Replace the access date when you cite it.
ScoreCompute. Bounded counterexample scan (scan_counterexample), version 0.2.0, source SHA-256 2df9bab83183ee8cf3f14cb995eaf60928279b792d8847a116959c28bfefcdf0. Release 0.16.1. https://scorecompute.com/engines/scan-counterexample/ (accessed YYYY-MM-DD).
All engine citations · Evidence card
Execution and availability
ScoreCompute exposes this tool through MCP Streamable HTTP. Rust computation runs on a connected worker; the public website and MCP gateway run separately. Among these 22 scientific tools, CUDA is implemented for simulate_pi; the other engines currently run on CPU. The separate contributor network has its own fixed integer Monte Carlo workload. Requests are bounded and concurrent work may be refused when capacity is occupied.
Record inputs, assumptions and returned provenance when sharing a result. The public observatory displays software client names and tool activity, without publishing calculation arguments or results.