OpenSymbolicAI Claude Skills

Claude Code skills for building agents with the OpenSymbolicAI framework.

What is OpenSymbolicAI?

A Python framework for building LLM-powered agents as deterministic software. Stop prompting. Start programming.

Three blueprint patterns:

• PlanExecute — Single-turn: LLM plans, Python executes
• DesignExecute — Adds control flow (if/for/while/try)
• GoalSeeking — Iterative plan-execute-evaluate loops

Skills Included

Skill	Command	Description
opensymbolicai	(auto-loaded)	API reference and framework context
create-agent	/create-agent	Scaffold a new agent project
add-primitive	/add-primitive	Add a @primitive method
add-decomposition	/add-decomposition	Add a @decomposition example
add-evaluator	/add-evaluator	Add an @evaluator to a GoalSeeking agent
debug-trace	/debug-trace	Debug execution traces and failures

Installation

Option 1: Install script (recommended)

curl -fsSL https://raw.githubusercontent.com/OpenSymbolicAI/claude-skills/main/install.sh | bash

Option 2: Clone directly

git clone https://github.com/OpenSymbolicAI/claude-skills.git /tmp/claude-skills
mkdir -p .claude/skills
cp -r /tmp/claude-skills/skills/* .claude/skills/
rm -rf /tmp/claude-skills

Option 3: Per-user (all projects)

git clone https://github.com/OpenSymbolicAI/claude-skills.git /tmp/claude-skills
mkdir -p ~/.claude/skills
cp -r /tmp/claude-skills/skills/* ~/.claude/skills/
rm -rf /tmp/claude-skills

Requirements

• Claude Code CLI
• Python 3.12+
• opensymbolicai-core package

Ecosystem

• core-py — Framework library
• examples-py — Example agents (date, RAG, deep research, multi-hop, unit converter)
• cli-py — Interactive TUI for running agents
• benchmark-py-MultiHopRAG — Multi-hop QA benchmark
• benchmark-py-folio — First-order logic benchmark
• core-ts — TypeScript core
• examples-ts — TypeScript examples