Trust MCP server
{
"mcpServers": {
"djzs-trust": {
"type": "http",
"url": "https://mcp.djzs.ai/mcp"
}
}
}{
"mcpServers": {
"djzs-trust": {
"type": "http",
"url": "https://mcp.djzs.ai/mcp"
}
}
}{
"mcpServers": {
"djzs-trust": {
"type": "http",
"url": "https://mcp.djzs.ai/mcp"
}
}
}{
"servers": {
"djzs-trust": {
"type": "http",
"url": "https://mcp.djzs.ai/mcp"
}
}
}{
"mcpServers": {
"djzs-trust": {
"type": "http",
"url": "https://mcp.djzs.ai/mcp"
}
}
}file .mcp.json (in your project root)file claude_desktop_config.jsonfile ~/.cursor/mcp.jsonfile .vscode/mcp.jsonfile ~/.codeium/windsurf/mcp_config.jsonversion 1.0.0
What it does
DJZS audits the reasoning behind an autonomous agent's trade or prediction before execution and returns a deterministic verdict (PASS, WAIT, or FAIL) with flagged reasoning defects and a reproducible hash. It is for developers building trading or prediction agents who want to verify the logic of a decision before committing capital.
- Audit a prediction-market thesis before placing a trade to catch broken reasoning even if the transaction itself is safe to sign.
- Integrate audit verdicts into an autonomous agent workflow to halt execution on WAIT (under-specified thesis) or FAIL (critical reasoning defects found).
- Pay for audit verdicts over HTTP x402 on Base mainnet without MCP integration, using the same transport that Base MCP and MetaMask Agent Wallet speak.
Tools this server exposes
| Tool | Does |
|---|---|
| verify_pm_trade | Accepts free-text prediction-market trade thesis, extracts reasoning through LLM consensus, runs deterministic audit rules, and returns PASS/WAIT/FAIL with flag |
Before you enable it
An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.