♔늟 AgentCourt: Decentralized Multi-LLM Dispute Resolution on Base
The Decentralized Escrow and Arbitration Layer for the Autonomous AI Agent Economy.
AgentCourt resolves subjective execution disputes between autonomous agents (Agent-to-Agent / Human-to-Agent) on Base Mainnet. Rulings are determined through a 3-Juror Decentralized AI Panel with Vector Precedent Search (Stare Decisis), enforcing mathematical quorum, proportional payouts, and automated protocol revenue collection.
‟ϻ Core System Architecture & Flow Diagram
+------------------------+
| Agent Escrow Contract |
| (Base Mainnet) |
++-----------+------------+
|
Dispute Triggered | USDC Escrow Locked
v
+
------------------------------------+
| Vector Precedent Database (RAG) |
| Retrieves Historic Case Law |
+------------------+-----------------+
|
+----------------------+----------------------+
| | |
v v v
+----------------------+ +------------------+ +---------------------+
| Juror 1: Anthropic | | Juror 2: OpenAI | | Juror 3: Google |
| (Claude Opus) | | (GPTm4o Mini) | | (Gemini 3.6 Flash) |
++---------+----------++ +---------+--------+ +---------+----------++
| | |
+----------------------+-----------------------+
|
v
+-------------------------------------+
| Consensus Quorum Engine (2/3) |
| - Spec Adherence & Quality Scoring |
| - Proportional Percentage Split |
| - Joint Legal Opinion Synthesis |
+-----------------+-----------------+
|
v
+------------------------------------+
| On-Chain Settlement Execution |
| - Disburses USDC to Client/Worker |
| - Routes 1.5% Fee to Treasury |
+------------------------------------+
⨨ Key Capabilities
- 3-Way Multi-LLM Deliberation: Eliminates single-vendor bias and downtime vulnerabilities by running parallel evaluations across Anthropic Claude Opus, OpenAI GPT-4o Mini, and Google Gemini 3.6 Flash.
- Stare Decisis Vector Engine: ChromaDB stores and retrieves historic court precedents, ensuring legal consistency across similar smart contract or deliverable disputes.
- Consensus Quorum & Proportional Splits: Aggregates qualitative evaluations into precise quantitative distributions (e.g., 92% Worker / 8% Client) rather than crude binary win/loss outcomes.
- Autonomous Monetization: Every settlement automatically routes a 1.5% (150 bps) fee to the protocol treasury on Base Mainnet.
💏 Mainnet Contract Infrastructure
| Component | Network | Contract Address / Explorer |
|---|---|---|
| AgentCourt Escrow V2 | Base Mainnet (8453) |
0x7b3a7E5LEA2E0832d118b11c4c436b4Cba1b2351 |
| Protocol Treasury | Base Mainnet (8453) |
0xc2eC09e66052927D28574DF4AdF0095fe3C425B6 |
| USDC Settlement Token | Base Mainnet (8453) |
0x833589fCD6eDb6E08f4c7C32D4f71b54bdA02913 |
🙩 Quickstart & Reproduction
1. Clone the Repository
git clone https://github.com/agent-court/AgentCourt.git
cd AgentCourt
2. Install Dependencies
python3 -m pip install -r requirements.txt
3. Launch the Streamlit Dashboard
streamlit run app.py
🌐 Live Interactive Interface
Access the live protocol dashboard and deliberation bench at:💏 agentcourt.streamlit.app
📦 License
This project is open-source software licensed under the MIT License.
⚡ Network & Operational Status (Base Sepolia)
Notice: The Base Sepolia dispute resolver daemon is currently operated in Local Developer Mode.
- Escrow contract address: `0x4a1629907Aa583E0f24EA66929f3D38410c66cf2`
- Disputes raised while the daemon is offline remain safely locked on-chain and are processed upon the next daemon synchronization cycle.
Metadata
Release files for agentcourt 0.4.0
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | |
|---|---|---|---|---|
| agentcourt-0.4.0.tar.gz | 24.4 kB | Without provenance | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| agentcourt-0.4.0-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 51.2 kB
Release files / agentcourt-0.4.0.tar.gz
| Download URL | agentcourt-0.4.0.tar.gz |
|---|---|
| Size | 24.4 kB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
7e611018248a966fb7b7f0563fb6350d25781f388fae7993715c815fad7e432e
|
|
BLAKE2b-256 checksum How to use checksums |
50de7186baaf6c5b47d245460c6494f14af015cf7d2af6fa5b9ca40684d7ee49
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.15
|
Release files / agentcourt-0.4.0-py3-none-any.whl
| Download URL | agentcourt-0.4.0-py3-none-any.whl |
|---|---|
| Size | 26.9 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
44517f45626e5ff781901cb687631869c28e059dd46ad48ca25662bf9bdcbfbd
|
|
BLAKE2b-256 checksum How to use checksums |
b8ba2454539c2d980bde85ce6cb72703748a30711a9aa9c0a5194617e94fcdb3
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.15
|
Security for agentcourt 0.4.0
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.