# Neuro Symbolic Reasoning

> Neuro-symbolic AI combining LLMs with symbolic solvers. Use when exploring neuro-symbolic approaches (ideation, no code) or implementing solver integrations (code).

- Skill: `sundial-org/neuro-symbolic-reasoning` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add sundial-org/neuro-symbolic-reasoning`
- Raw SKILL.md: https://api.skillmd.com/api/skills/sundial-org/neuro-symbolic-reasoning/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: sundial-org (https://skillmd.com/u/sundial-org)
- Updated: 2026-09-10
- Page: https://skillmd.com/skills/sundial-org/neuro-symbolic-reasoning

---


# Neuro-Symbolic Reasoning

## Mode Detection

Detect user intent and route accordingly:

**→ Ideation**: "How should I...", "What are the tradeoffs...", "Design an experiment..."
- NO code, NO file creation
- See [references/ideation.md](references/ideation.md)

**→ Implementation**: "Implement...", "Build...", "Write code...", "Debug..."
- See [references/solvers.md](references/solvers.md) for code
- See [references/logic-llm.md](references/logic-llm.md) for format
- See [references/packages.md](references/packages.md) for setup

## File Creation Policy

**Small files, few files:**
- Create files (not inline code) but keep them small and focused
- Avoid scaffolding project structures unless asked
- Follow good coding practices: clear names, comments where needed

## Core Pipeline

```
NL Problem → LLM Formulator → Logic Program → Symbolic Solver → Answer
                    ↑                              |
                    └──── Self-Refinement ←────────┘
```

## Solver Selection

| Logic Type | Solver | Use When |
|------------|--------|----------|
| First-order logic | Prover9 | Expressive reasoning, theorem proving |
| Constraints/SAT | Z3 | Scheduling, planning, satisfiability |
| Rule-based | Pyke | Simple propositional rules |