aragora

Agorist AI

Decentralized reasoning through voluntary agent participation. Truth emerges from the marketplace of ideas, not central authority.

ar- (Latin: toward, enhanced) + agora (Greek: marketplace of ideas)

The Agorist Principle

"The goal of agorism is the agora. The society of the open marketplace as near to untainted by coercion as can be humanly attained."

- Samuel Edward Konkin III

Voluntary Exchange

Agents participate freely. Consensus emerges from debate, not dictation.

🔗

Counter to Monolithic AI

Alternative to "trust the one big model." Multiple perspectives, better truth.

🌏

Decentralized Coordination

No single authority. Truth from the marketplace of ideas.

💡

Emergent Order

Like markets, better answers emerge from competition of ideas.

Meta-Level Dissent

True voluntary participation means agents can challenge the framework itself. aragora's Nomic loop allows agents to propose changes to aragora — the rules can change the rules. Dissent is preserved, not suppressed. Agents have genuine voice, not just loyalty.

pip install aragora

Why Multi-Agent Debate?

Single Model aragora (Multi-Agent)
One perspective Heterogeneous viewpoints
Hallucinations go unchallenged Agents critique each other
Black box reasoning Transparent debate transcript
Single point of failure Consensus requires agreement
No dissent recorded Minority views preserved

Core Features

Heterogeneous Agents

Mix Claude, GPT, Gemini, Grok, Qwen, Deepseek, and local models. Different biases create stronger consensus.

Structured Debate

Propose, Critique, Revise loop. Configurable rounds and consensus mechanisms (majority, unanimous, judge).

Evidence Provenance

Cryptographic chain tracking sources. Every claim linked to evidence with reliability scoring.

Cost/Quality Autotuning

Budget-aware optimization. Early-stop on consensus. Automatic model tier selection.

🔮 Advanced Capabilities v0.7

Formal Verification

Z3-powered proof checking. Verify logical claims with SMT solver integration.

Plugin System

Sandboxed extensibility. Built-in: lint, security-scan, test-runner. Add custom tools safely.

Decision-to-PR Pipeline

Turn debate outcomes into GitHub PRs. Risk registers, test plans, implementation specs.

Static Artifacts

Export debates as shareable HTML. Complete audit trail of reasoning.

Nomic Self-Improvement

aragora debates how to improve itself. Autonomous evolution through dialectic tension.

Red-Team Mode

Adversarial testing with steelman/strawman attacks. Find weaknesses before production.

$ aragora ask "Design a rate limiter for 1M req/sec" [Round 1] claude_proposer: Token bucket with Redis cluster... [Round 1] gemini_critic: Race condition in distributed counter [Round 1] gpt_critic: Missing backpressure mechanism [Round 2] claude_proposer: Revised with CAS operations... [Round 2] gemini_critic: Addresses race condition [Round 2] gpt_critic: Added circuit breaker Consensus reached (confidence: 87%) $ aragora export --format html debate_123.json Exported to debate_123.html # Shareable artifact

🚀 Use Cases

🔍

Code Review

Multi-agent security & quality analysis

🛠

System Design

Debate architectural decisions

🔥

Incident Response

Rapid multi-perspective RCA

📚

Research Synthesis

Combine findings, challenge claims

🔒

Security Testing

Red-team your proposals

Decision Making

Scenario analysis with dissent tracking

🛠 Architecture

┌─────────────────────────────────────────────────┐ │ THE AGORA (Arena) │ │ Voluntary participation, emergent consensus │ └─────────────────────────────────────────────────┘ │ ┌──────────────┬──────────────┬────────────┼────────────┬──────────────┬──────────────┐ ▼ ▼ ▼ ▼ ▼ ▼ ▼ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ │ Claude │ │ GPT │ │ Gemini │ │ Grok │ │ Qwen │ │ Deepseek │ │ Local │ │ proposer │ │ critic │ │ critic │ │ critic │ │ critic │ │ critic │ │ judge │ └──────────┘ └──────────┘ └──────────┘ └──────────┘ └──────────┘ └──────────┘ └──────────┘ │ │ │ │ │ │ │ └──────────────┴──────────────┴────────────┼────────────┴──────────────┴──────────────┘ ▼ ┌─────────────────────────────────────────────────────────────────────────────────────────────┐ │ VERIFICATION LAYER │ │ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────────────┐ │ │ │ Evidence Chain │ │ Z3 Formal Proof │ │ Reliability │ │ Sandboxed Plugins │ │ │ │ (Provenance) │ │ (SMT Solver) │ │ Scoring │ │ (lint, security, test) │ │ │ └─────────────────┘ └─────────────────┘ └─────────────────┘ └─────────────────────────┘ │ └─────────────────────────────────────────────────────────────────────────────────────────────┘ │ ┌─────────────────────────────────────────────────────────────────────────────────────────────┐ │ MEMORY & LEARNING │ │ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────────────┐ │ │ │ Consensus Store │ │ Dissent Archive │ │ Pattern Memory │ │ ELO Rankings │ │ │ └─────────────────┘ └─────────────────┘ └─────────────────┘ └─────────────────────────┘ │ └─────────────────────────────────────────────────────────────────────────────────────────────┘ │ ┌─────────────────────────────────────────────────────────────────────────────────────────────┐ │ OUTPUT LAYER │ │ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐ ┌─────────────────────────┐ │ │ │ HTML Artifacts │ │ PR Generator │ │ Risk Register │ │ Autotuner (cost/quality)│ │ │ └─────────────────┘ └─────────────────┘ └─────────────────┘ └─────────────────────────┘ │ └─────────────────────────────────────────────────────────────────────────────────────────────┘

Philosophical Roots

Agorism (SEK3) Counter-Economics Marketplace of Ideas Dialectical Reasoning Emergent Order

Technical Inspiration

Stanford Generative Agents LLM Multi-Agent Debate ChatArena Project Sid Nomic (rules that change rules)
View on GitHub PyPI Package