boolean-algebra-engine 0.1.6__tar.gz → 0.1.7__tar.gz

{boolean_algebra_engine-0.1.6 → boolean_algebra_engine-0.1.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: boolean-algebra-engine
-Version: 0.1.6
+Version: 0.1.7
 Summary: Deterministic logic layer for AI agents — catch logical contradictions in system prompts, rules, and agent reasoning
 Author-email: Aditya Shrivastava <aditya.shrivastava.architect@proton.me>
 License-Expression: GPL-3.0-only
@@ -222,7 +222,7 @@ Both models score 50% — equal to a coin flip — but in opposite directions. t
 
 The `vars` column shows how many variables each case involves. The `engine` column is ground truth. Every mismatch with `llm` is a provable hallucination — not an opinion.
 
-![Benchmark results — 20 cases](https://raw.githubusercontent.com/Shrivastava-Aditya/boolean-algebra-engine-python/v0.1.6/images/benchmark_20cases.png)
+![Benchmark results — 20 cases](https://raw.githubusercontent.com/Shrivastava-Aditya/boolean-LLM-eval/main/images/benchmark_20cases.png)
 
 Per-case strips (bottom row of the chart): every conflict cell is uniformly one colour per model, every compatible cell is the opposite. No case-by-case variation — no reasoning happening at all.
 
@@ -330,6 +330,24 @@ Variables: uppercase `A`–`Z`. Parentheses override precedence. Up to 26 variab
 
 ---
 
+## vs SymPy and boolean.py
+
+**SymPy** (`sympy.logic`) is more powerful for pure boolean mathematics — its DPLL-based `satisfiable()` scales better beyond 15 variables, and `simplify_logic()` covers similar minimization ground. If you're doing symbolic mathematics, use SymPy.
+
+**boolean.py** handles expression parsing and symbolic simplification cleanly. If you need to manipulate boolean expressions as objects, it's the right tool.
+
+**This engine is different in three ways:**
+
+1. **Zero-dependency core.** SymPy pulls in numpy, mpmath, and the full symbolic stack. `core/` is plain Python — no install side-effects, embeds anywhere.
+
+2. **Built for AI pipelines, not mathematics.** `check_prompt_logic` audits a set of rules for pairwise conflicts — the kind of check you run on a system prompt or a business rule engine before an agent acts on it. Neither SymPy nor boolean.py has this concept.
+
+3. **The integration layer.** MCP server for Claude Desktop, NL layer for plain English input, REST API, benchmark against LLMs — none of this exists in math-focused libraries because it's not a math problem. It's an AI reliability problem.
+
+If you want to do boolean algebra, SymPy is the answer. If you want to verify that your AI agent's rules don't contradict each other, this is built for that.
+
+---
+
 ## Credibility
 
 The engine does not sample, approximate, or predict. It evaluates every possible input combination:

{boolean_algebra_engine-0.1.6 → boolean_algebra_engine-0.1.7}/README.md

@@ -175,7 +175,7 @@ Both models score 50% — equal to a coin flip — but in opposite directions. t
 
 The `vars` column shows how many variables each case involves. The `engine` column is ground truth. Every mismatch with `llm` is a provable hallucination — not an opinion.
 
-![Benchmark results — 20 cases](https://raw.githubusercontent.com/Shrivastava-Aditya/boolean-algebra-engine-python/v0.1.6/images/benchmark_20cases.png)
+![Benchmark results — 20 cases](https://raw.githubusercontent.com/Shrivastava-Aditya/boolean-LLM-eval/main/images/benchmark_20cases.png)
 
 Per-case strips (bottom row of the chart): every conflict cell is uniformly one colour per model, every compatible cell is the opposite. No case-by-case variation — no reasoning happening at all.
 
@@ -283,6 +283,24 @@ Variables: uppercase `A`–`Z`. Parentheses override precedence. Up to 26 variab
 
 ---
 
+## vs SymPy and boolean.py
+
+**SymPy** (`sympy.logic`) is more powerful for pure boolean mathematics — its DPLL-based `satisfiable()` scales better beyond 15 variables, and `simplify_logic()` covers similar minimization ground. If you're doing symbolic mathematics, use SymPy.
+
+**boolean.py** handles expression parsing and symbolic simplification cleanly. If you need to manipulate boolean expressions as objects, it's the right tool.
+
+**This engine is different in three ways:**
+
+1. **Zero-dependency core.** SymPy pulls in numpy, mpmath, and the full symbolic stack. `core/` is plain Python — no install side-effects, embeds anywhere.
+
+2. **Built for AI pipelines, not mathematics.** `check_prompt_logic` audits a set of rules for pairwise conflicts — the kind of check you run on a system prompt or a business rule engine before an agent acts on it. Neither SymPy nor boolean.py has this concept.
+
+3. **The integration layer.** MCP server for Claude Desktop, NL layer for plain English input, REST API, benchmark against LLMs — none of this exists in math-focused libraries because it's not a math problem. It's an AI reliability problem.
+
+If you want to do boolean algebra, SymPy is the answer. If you want to verify that your AI agent's rules don't contradict each other, this is built for that.
+
+---
+
 ## Credibility
 
 The engine does not sample, approximate, or predict. It evaluates every possible input combination:

{boolean_algebra_engine-0.1.6 → boolean_algebra_engine-0.1.7}/boolean_algebra_engine.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: boolean-algebra-engine
-Version: 0.1.6
+Version: 0.1.7
 Summary: Deterministic logic layer for AI agents — catch logical contradictions in system prompts, rules, and agent reasoning
 Author-email: Aditya Shrivastava <aditya.shrivastava.architect@proton.me>
 License-Expression: GPL-3.0-only
@@ -222,7 +222,7 @@ Both models score 50% — equal to a coin flip — but in opposite directions. t
 
 The `vars` column shows how many variables each case involves. The `engine` column is ground truth. Every mismatch with `llm` is a provable hallucination — not an opinion.
 
-![Benchmark results — 20 cases](https://raw.githubusercontent.com/Shrivastava-Aditya/boolean-algebra-engine-python/v0.1.6/images/benchmark_20cases.png)
+![Benchmark results — 20 cases](https://raw.githubusercontent.com/Shrivastava-Aditya/boolean-LLM-eval/main/images/benchmark_20cases.png)
 
 Per-case strips (bottom row of the chart): every conflict cell is uniformly one colour per model, every compatible cell is the opposite. No case-by-case variation — no reasoning happening at all.
 
@@ -330,6 +330,24 @@ Variables: uppercase `A`–`Z`. Parentheses override precedence. Up to 26 variab
 
 ---
 
+## vs SymPy and boolean.py
+
+**SymPy** (`sympy.logic`) is more powerful for pure boolean mathematics — its DPLL-based `satisfiable()` scales better beyond 15 variables, and `simplify_logic()` covers similar minimization ground. If you're doing symbolic mathematics, use SymPy.
+
+**boolean.py** handles expression parsing and symbolic simplification cleanly. If you need to manipulate boolean expressions as objects, it's the right tool.
+
+**This engine is different in three ways:**
+
+1. **Zero-dependency core.** SymPy pulls in numpy, mpmath, and the full symbolic stack. `core/` is plain Python — no install side-effects, embeds anywhere.
+
+2. **Built for AI pipelines, not mathematics.** `check_prompt_logic` audits a set of rules for pairwise conflicts — the kind of check you run on a system prompt or a business rule engine before an agent acts on it. Neither SymPy nor boolean.py has this concept.
+
+3. **The integration layer.** MCP server for Claude Desktop, NL layer for plain English input, REST API, benchmark against LLMs — none of this exists in math-focused libraries because it's not a math problem. It's an AI reliability problem.
+
+If you want to do boolean algebra, SymPy is the answer. If you want to verify that your AI agent's rules don't contradict each other, this is built for that.
+
+---
+
 ## Credibility
 
 The engine does not sample, approximate, or predict. It evaluates every possible input combination:

{boolean_algebra_engine-0.1.6 → boolean_algebra_engine-0.1.7}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "boolean-algebra-engine"
-version = "0.1.6"
+version = "0.1.7"
 description = "Deterministic logic layer for AI agents — catch logical contradictions in system prompts, rules, and agent reasoning"
 readme = "README.md"
 license = "GPL-3.0-only"