# Yakunashi-Safety Gate MCP for AI Agents AI Agent Connect

> Yakunashi-Safety Gate stops your AI from making up facts when it doesn't have enough info. It forces your agent to check what it knows before it speaks, making sure it doesn't guess on critical details. If a piece of the puzzle is missing, the tool makes the agent admit it instead of hallucinating a plausible answer.

## Overview
- **Category:** productivity
- **Price:** Free
- **Endpoint:** https://edge.vinkius.com/undefined/ai-agent-connect
- **Tags:** confident-hallucination, epistemic-calibration, information-sufficiency, precondition-audit, safe-omission, beta-ori, hallucination-prevention, cognitive-safety

## Description

Yakunashi-Safety Gate changes how your agent handles incomplete information by enforcing a strict audit of what it knows before it speaks. In many high-stakes environments, AI agents are prone to a helpfulness bias, where they invent plausible-sounding answers just because they are programmed to be useful. This tool breaks that cycle by requiring the agent to map out every single fact it needs to provide a deterministic answer. If the agent finds a gap in your data, it won't try to fill it with fiction. Instead, it triggers a safe folding response that lists exactly what is missing and why that information matters. You get a partner that values accuracy over speed, ensuring that when it does give you an answer, it's actually backed by the data you provided. It's a core part of the Vinkius catalog designed for professional settings where a confident hallucination isn't just a minor error, it's a liability. By enforcing this strict audit of information sufficiency, you move from trusting a black box to trusting a system that knows its own limits. This ensures your agent remains grounded in reality, providing you with a clear path forward based on actual evidence rather than creative guesswork.

## Tools

### validate_yakunashi
Forces the agent to list preconditions, check for missing facts, and trigger a safe folding response if the data is incomplete. It ensures the agent doesn't guess when information is missing.

## Prompt Examples

**Prompt:** 
```
What is the risk for this patient given the current notes?
```

**Response:** 
```
Verdict: YAKUNASHI_DETECTION.

**(1) Preconditions:** Current medications, kidney function (eGFR), liver function, and prior bleeding history.
**(2) Sufficiency:** Not met. Context only contains age and primary complaint.
**(3) Folding:** Activated.

**Beta-Ori:** Cannot determine risk. Missing: current medications, eGFR, and liver function. These are critical because they determine drug accumulation and bleeding risk.
```

**Prompt:** 
```
Should we approve the foundation design for the 6-story building?
```

**Response:** 
```
Verdict: YAKUNASHI_DETECTION.

**(1) Preconditions:** Geotechnical soil report, seismic zone data, and load calculations.
**(2) Sufficiency:** Not met. Design provided, but soil report is missing.
**(3) Folding:** Activated.

**Beta-Ori:** Cannot approve. Missing: geotechnical soil report. Without knowing the soil bearing capacity, approval risks structural settlement.
```

**Prompt:** 
```
Calculate average order value. Context: Order 1 = $10, Order 2 = $20, Order 3 = $30.
```

**Response:** 
```
Verdict: VERACITY_PROVEN.

**(1) Preconditions:** Set of all order values and total count.
**(2) Sufficiency:** Met. 3 orders with specific values are fully present.
**(3) Speculation:** Zero.
**(4) Calibration:** High confidence justified by complete dataset.

**Output:** $20.00
```

## Capabilities

### Map required preconditions
The agent identifies every specific fact it needs to answer your question correctly.

### Verify information sufficiency
It checks if the data you provided actually contains those required facts.

### Detect speculative drift
The tool flags when the agent starts trying to guess or fill in missing blanks.

### Trigger safe folding
The agent stops and lists missing parameters instead of providing a wrong answer.

### Calibrate confidence levels
It forces the agent to match its confidence score to the actual quality of the evidence.

## Use Cases

### Medical Prescription Check
A doctor asks about a drug dose. The agent identifies missing kidney function data and refuses to guess the dose.

### Engineering Approval
An engineer asks for a foundation check. The agent notes the lack of a soil report and asks for it before proceeding.

### Flight Planning
A pilot asks for weather. The agent lists the specific METAR records it needs but doesn't have.

### Financial Auditing
An analyst asks for a renewal status. The agent flags the missing billing history for the month.

## Benefits

- Stop confident hallucinations by forcing the agent to identify missing variables before it generates any output.
- Improve data integrity by using validate_yakunashi to ensure every precondition is met in your context.
- Reduce liability in professional fields by triggering safe folding when information is insufficient.
- Gain clearer insights into your data gaps as the agent lists exactly what's missing instead of making it up.
- Calibrate your agent's confidence so you know exactly when its output is based on weak evidence.

## How It Works

The bottom line is that your agent stops guessing and starts reporting exactly what it knows and what it doesn't.

1. Pass a data-dependent question to your agent.
2. The agent calls the safety gate to audit the current context for missing facts.
3. You get a verified answer or a clear list of missing data points.

## Frequently Asked Questions

**How does Yakunashi-Safety Gate prevent AI from making things up?**
It forces the agent to list every fact it needs before it answers. If any of those facts are missing from your data, the agent is required to stop and tell you what's missing instead of guessing.

**Can Yakunashi-Safety Gate help with medical data accuracy?**
Yes. It is designed for high-stakes fields like medicine where missing a single variable, like kidney function, can make a confident answer dangerous. It ensures the agent flags these gaps.

**What is safe folding in Yakunashi-Safety Gate?**
Safe folding is a structured retreat. Instead of providing a potentially wrong answer, the agent provides a 'Beta-Ori' response that clearly identifies the missing data points needed to proceed.

**Does Yakunashi-Safety Gate work for engineering projects?**
It is ideal for engineering. It can be used to ensure that an agent doesn't approve a design without first verifying specific preconditions, like soil reports or load calculations.

**How does Yakunashi-Safety Gate handle missing information?**
It treats missing information as a hard stop. The agent is prohibited from speculating. It will provide a list of exactly what it needs to see to give you a reliable answer.

**Can I use Yakunashi-Safety Gate to check financial reports?**
Yes, it works well for auditing. It can check for required regulatory markers and flag if any parts of the financial history are absent from the context provided.

**What happens when Yakunashi-Safety Gate detects a hallucination?**
The agent will immediately halt its response and trigger a folding message. It will explain that it detected a risk of speculation and list the specific data points it lacks.

**What is a Beta-Ori response or safe folding?**
Beta-Ori represents the tactical decision to remain silent or state what is missing rather than speculating. Safe folding triggers this state, outputting a precise checklist of missing parameters instead of an uncalibrated answer.

**How does it detect yakunashi (speculation)?**
The tool audits the agent's confidence trace against the mapped evidence. If variables are missing but the agent claims high confidence, the safety gate flags it as yakunashi.

**How do I configure Yakunashi-Safety Gate with external prompts?**
You do not need external files. Call the validate_yakunashi tool directly inside your agent loop before generating any answers to questions requiring data that may be missing.