Concordance Engine

@matharrismma

About Concordance Engine

O(1) authority-based validation across seven domains (chemistry, physics, mathematics, statistics, computer science, biology, governance). Replaces O(n²) consensus coordination. Four-gate pipeline: RED, FLOOR, BROTHERS, GOD. MCP server included — connects to Claude Desktop and Cl

Config

Add this server to your MCP-compatible client using the configuration below.

{
  "mcpServers": {
    "concordance-engine": {
      "command": "concordance-mcp"
    }
  }
}

Tools

No tools detected

We auto-extract tools from the README. The maintainer can list them under a ## Tools heading to populate this section.

Overview

What is Concordance Engine?

Concordance Engine is a Python validation engine and MCP server that checks decision packets and computational claims against fixed external authorities (conserved laws, proven mathematics, pre‑registered statistical methods) rather than polling internal consensus. It implements O(1) validation, halting at the first gate failure, and never self‑confirms. Designed for AI systems, researchers, and governance workflows.

How to use Concordance Engine?

Install with pip install -e ".[mcp]" then run concordance-mcp to start the MCP server. Configure in Claude Desktop or add via claude mcp add. Alternatively, use the CLI: concordance validate <packet.json>. MCP tools return {"status":"ERROR",…} on bad input, making them safe inside LLM tool loops.

Key features of Concordance Engine

O(1) external authority validation (replaces O(n²) consensus).
Eleven computational verifiers (chemistry, physics, math, statistics, CS, biology, governance).
Four‑gate pipeline (RED, FLOOR, BROTHERS, GOD) with deterministic halting.
Safe for LLM loops: tools never raise exceptions.
MCP integration with Claude Desktop and Claude Code.
CLI with summary/verbose/json output.

Use cases of Concordance Engine

Validate governance decision packets for structural completeness and witness consistency.
Verify chemistry equation balance, physics dimensional analysis, and conservation laws.
Recompute p‑values and multiple‑comparison corrections from raw statistical inputs.
Check computer‑science code for termination, functional correctness, and runtime complexity.
Enforce biological experiment validity (replicates, dose‑response, power analysis).

FAQ from Concordance Engine

How is Concordance Engine different from consensus‑based validation?

Consensus‑based validation scales O(n²) and depends on the least reliable participant. Concordance Engine validates every claim against fixed external standards, scaling O(1) independent of participant count.

What dependencies are required?

Required: sympy>=1.12, numpy>=1.26, scipy>=1.11. Optional: jsonschema>=4.21.0 (full schema validation), mcp>=1.0.0 (MCP server).

How do verifiers handle invalid input?

Each verifier returns a status (CONFIRMED, MISMATCH, ERROR, NOT_APPLICABLE) with a human‑readable detail string and structured data. They never raise exceptions.

Can I add custom verifiers or domains?

Yes. Add a domain validator in domains/<domain>.py exposing validate_red(packet) and validate_floor(packet), or add a computational verifier in verifiers/<domain>.py with a run(packet) function and register it in verifiers/__init__.py.

What are the four gates?

RED (hard, attestation + computational verification → REJECT), FLOOR (hard, structural rules → REJECT), BROTHERS (soft, witness count → QUARANTINE), GOD (soft, elapsed wait → QUARANTINE). The engine halts at the first gate failure.

Frequently asked questions

How is Concordance Engine different from consensus‑based validation?

What dependencies are required?

Required: `sympy>=1.12`, `numpy>=1.26`, `scipy>=1.11`. Optional: `jsonschema>=4.21.0` (full schema validation), `mcp>=1.0.0` (MCP server).

How do verifiers handle invalid input?

Each verifier returns a status (`CONFIRMED`, `MISMATCH`, `ERROR`, `NOT_APPLICABLE`) with a human‑readable `detail` string and structured `data`. They never raise exceptions.

Can I add custom verifiers or domains?

Yes. Add a domain validator in `domains/<domain>.py` exposing `validate_red(packet)` and `validate_floor(packet)`, or add a computational verifier in `verifiers/<domain>.py` with a `run(packet)` function and register it in `verifiers/__init__.py`.