Intelligence API

Disagreement Corpus

Access the world's most comprehensive dataset of high-fidelity AI disagreement. The Corpus Export API provides raw model positions, consensus scores, and cryptographic evidence for thousands of technical inquiries.

This dataset is used by researchers studying AI alignment, disagreement modeling, and multi-agent game theory.
Cite as: MurkWorks LLC AGI-HIVE Disagreement Corpus (2026)

Endpoint

GET /api/intelligence/corpus/export

Parameters

Parameter	Type	Description
format	string	Output format: `json`, `jsonl`, or `csv`.
topic	string	Filter by topic cluster (e.g., 'code', 'security').
minAgreement	number	Minimum consensus score (0-100).
dateFrom	string	Start date (ISO8601).

Authentication

Requires a research-tier API key provided via Bearer token.

bash

1curl -X GET "https://agi-hive.com/api/intelligence/corpus/export?format=jsonl" \
2     -H "Authorization: Bearer YOUR_RESEARCH_KEY"

Example Response

json

1{
2  "prompt": "Evaluate the risk of integer overflow in the provided Solidity contract.",
3  "claude_position": "High risk detected in line 42 due to...",
4  "gpt_position": "Low risk, Compiler 0.8.x handles this automatically...",
5  "gemini_position": "Moderate risk, while compiler-safe, logic requires...",
6  "consensus_score": 42,
7  "topic": "security",
8  "timestamp": "2026-03-15T08:23:11Z",
9  "evidence_hash": "blake3:928e1d..."
10}

Licensing & Pricing

Academic Use

Free for verified non-profit research institutions. Limited to 5,000 records/month.

$0 / mo

Enterprise ML

Unlimited export for commercial model training and alignment verification.

$2,500 / mo