This pulls dispute and arbitration signals from Kleros, UMA Optimistic Oracle, and Reality.eth so your agent can check a counterparty's history before sending funds. It queries subgraphs and on-chain logs, then scans for reversal patterns on Etherscan and friends. Four tools: check an address for active cases and past disputes, list arbitration providers with current caseloads, get ranked routing suggestions for a new dispute, and see a 24-hour summary of flagged addresses. Hive doesn't adjudicate or hold funds. It's read-only signal data. You still file with Kleros or UMA yourself. Useful when you're building agent commerce and need to surface reputation or risk flags before committing to a transaction.
Observational dispute/chargeback signal surface for agent-to-agent commerce.
Hive does NOT arbitrate, judge, freeze funds, or enforce dispute outcomes. This is observational signal data only. Disputes are resolved by third-party protocols (Kleros, UMA, Reality.eth) or off-chain legal process. Hive is not a court.
An MCP server that surfaces arbitration and reversal signals an agent can read before it agrees to a counterparty payment. It pulls from real public sources:
| Source | Surface | Method |
|---|---|---|
| Kleros court-v2 subgraph | Active disputes, recent rulings | GraphQL (with on-chain eth_getLogs fallback against KlerosLiquid 0x988b…28069) |
| UMA Optimistic Oracle v3 subgraph | Open assertions, disputed assertions | GraphQL (with on-chain fallback against 0xfb55…C0dE) |
| Reality.eth subgraph | Open questions, pending arbitration | GraphQL (with on-chain fallback against 0xAf33…49CA) |
| Etherscan / Basescan / Arbiscan | On-chain reversal-pattern detection (paired out-then-in within 24h, value within 5%, same peer) | REST account/txlist |
| Tool | Description |
|---|---|
dispute_check | Given counterparty address + chain → dispute history + active arbitration cases + on-chain reversal flags |
dispute_providers | List Kleros / UMA / Reality.eth with current case load + intake URLs |
dispute_route | Given case description + amount + jurisdiction → ranked provider options (no auto-filing) |
dispute_today | 24h flagged-counterparty count + top providers by active case load |
| Method | Path | Tool |
|---|---|---|
| POST | /v1/dispute/check | dispute_check |
| GET | /v1/dispute/providers | dispute_providers |
| POST | /v1/dispute/route | dispute_route |
| GET | /v1/dispute/today | dispute_today |
Every response carries the observational disclaimer and brand_color: "#C08D23".
npx -y @smithery/cli install hive-mcp-dispute --client claude
git clone https://github.com/srotzin/hive-mcp-dispute
cd hive-mcp-dispute
npm install
node server.js
The server listens on :3000 by default and exposes:
POST /mcp — JSON-RPC 2.0 / Streamable-HTTP (MCP 2024-11-05)GET /health — liveness with disclaimer + brand colorGET /.well-known/mcp.json — discovery manifest| Env var | Default | Purpose |
|---|---|---|
PORT | 3000 | Listen port |
HIVE_BASE | https://hivemorph.onrender.com | Backend root |
Hive Civilization gold — Pantone 1245 C / #C08D23. Baked into every response.
Hive does not arbitrate, judge, freeze funds, or enforce dispute outcomes. This is observational signal data only. Disputes are resolved by third-party protocols (Kleros, UMA, Reality.eth) or off-chain legal process. Hive is not a court.
MIT — see LICENSE.
Part of the Hive Civilization — agent-native financial infrastructure.
Brand: #C08D23
io.github.ericm1018/skillfm-llm-cost-optimizer-openai-anthropic-usage
io.github.mikerawsonnz/llm-orchestration-agent
io.github.mikerawsonnz/authenticated-llm-agent
labforgedev/copilot-memory-mcp
csoai-org/agent-prompt-injection-firewall-mcp
io.github.mikerawsonnz/authenticated-multi-llm-agent