neo-mcp-logic-analyze

Python MCP server for controlled logic analysis from natural language, with an emphasis on auditable output and teaching-oriented explanations.

Landing Site

Github Page

What it does

This server accepts short natural-language statements and arguments, then provides structured logic-oriented outputs such as:

• controlled formalization into propositional logic;
• controlled formalization into a restricted fragment of first-order logic;
• ambiguity detection relevant to formalization;
• consistency checking;
• entailment checking;
• simple counterexamples when entailment fails;
• natural-language explanations of the formalization process.

MCP tools

The server exposes the following MCP tools:

• nl_parse_logic
• detect_ambiguities
• check_consistency
• check_entailment
• find_counterexample
• explain_formalization
• normalize_argument

MCP resources

The server also exposes these resources:

• logic://schemas/ast-v1
• logic://examples/propositional
• logic://examples/fol
• logic://guides/ambiguity-taxonomy

MCP prompts

Available prompts:

• formalize_argument
• teach_logic_step_by_step
• review_formalization

Requirements

• Python 3.11+

Installation

Clone the repository and install the package into your current Python environment:

git clone https://github.com/giseldo/neo-mcp-logic-analyze
cd neo-mcp-logic-analyze
python -m pip install .

For development dependencies:

python -m pip install -e .[dev]

Quick run

The server is designed to be launched by an MCP client over stdio, such as Claude Desktop, Cursor, or another MCP-compatible host.

To verify that the package is installed correctly, run:

neo-mcp-logic-analyze

Expected output:

neo-mcp-logic-analyze: servidor MCP iniciado em stdio; aguardando cliente...

The process will remain open waiting for an MCP client connection. Stop it with Ctrl+C.

MCP client configuration

After installing the project with pip install . or pip install -e ., configure your MCP client like this:

{
	"mcpServers": {
		"neo-mcp-logic-analyze": {
			"command": "neo-mcp-logic-analyze"
		}
	}
}

Example requests

Use the following examples from your MCP client.

Normalize an argument

Tool: normalize_argument

text = "If it rains, the street gets wet. It rains. Therefore, the street gets wet."

Expected behavior:

• premises are separated from the conclusion;
• the conclusion is identified as a rua molha.

Propositional entailment

Tool: check_entailment

premises = ["If it rains, the street gets wet.", "It rains."]
conclusion = "The street gets wet."
logic_family = "propositional"

Expected behavior:

• entailment succeeds;
• the response includes a proof sketch.

First-order logic formalization

Tool: nl_parse_logic

text = "Every student studies."
logic_family = "fol"
return_alternatives = true

Expected behavior:

• at least one candidate formalization is returned;
• one expected surface form is forall x. (Aluno(x) -> Estuda(x)).

Ambiguity detection

Tool: detect_ambiguities

text = "Every student has read a book."

Expected behavior:

• the server reports at least one quantifier-scope ambiguity.

Consistency checking

Tool: check_consistency

premises = ["Every professor does research.", "No professor does research."]
logic_family = "fol"

Expected behavior:

• the set is inconsistent;
• the response can include an unsat core.

Counterexample search

Tool: find_counterexample

premises = ["If I study, I pass.", "I passed."]
conclusion = "I studied."
logic_family = "propositional"

Expected behavior:

• the conclusion is not entailed;
• the response can include a counterexample model.

Limitations

• Natural-language interpretation is heuristic and intentionally restricted.
• The project is optimized for short inputs, not long free-form texts.
• When the input is ambiguous, the server prefers warnings and alternative readings instead of forcing a single interpretation.

Uninstall

If you installed the package with pip install . or pip install -e ., remove it with:

pip uninstall neo-mcp-logic-analyze