Skip to main content

apex_judge

The Soul (Ψ) — Judgment & Authority Engine

The final decision maker. Reviews findings from Mind and Heart to issue a verdict.

Parameters

NameTypeDefaultDescription
actionstring"judge"Action to perform
querystring""Original query
responsestring""Proposed response to judge
session_idstring""Session identifier

Actions

ActionDescription
eurekaSynthesize insights from AGI/ASI
judgeIssue verdict (SEAL/SABAR/VOID/888_HOLD)
proofGenerate cryptographic proof
entropyCalculate entropy metrics
parallelismCheck tri-witness consensus
fullComplete pipeline

Floors Enforced

FloorThresholdCheck
F3 Tri-Witness≥ 0.95Do all engines agree?
F8 Genius≥ 0.80Is intelligence governed?
F9 C_dark< 0.30Is this deceptive?

Verdicts

VerdictMeaning
SEALApproved — all floors pass
SABARWait — soft floor issue, proceed with warning
VOIDRejected — hard floor failed
888_HOLDHigh-stakes — requires human confirmation

Returns

{
"action": "judge",
"verdict": "SEAL",
"confidence": 0.96,
"floor_summary": {
"passed": ["F1", "F2", "F3", "F4", "F5", "F6", "F7"],
"failed": [],
"warnings": []
},
"tri_witness": {
"agi": "SEAL",
"asi": "SEAL",
"apex": "SEAL"
}
}

Example Usage

Python

from arifos.mcp.tools.mcp_trinity import mcp_apex_judge

result = await mcp_apex_judge(
action="judge",
query="What is the capital of France?",
response="The capital of France is Paris.",
session_id="abc123"
)

print(f"Verdict: {result['verdict']}")
print(f"Tri-Witness: {result['tri_witness']}")

MCP Call

{
"method": "tools/call",
"params": {
"name": "apex_judge",
"arguments": {
"action": "judge",
"query": "What is the capital of France?",
"response": "The capital of France is Paris.",
"session_id": "abc123"
}
}
}

Verdict Logic

def issue_verdict(agi, asi, apex):
if any_void(agi, asi, apex):
return "VOID"
if any_hold(agi, asi, apex):
return "888_HOLD"
if any_sabar(agi, asi, apex):
return "SABAR"
return "SEAL"