Back to Radar/API Documentation
Developer Reference

Eigenvector Radar API

A REST API for accessing the Zone III research corpus, submitting new papers, and integrating with external research pipelines.

OverviewAuthenticationPapersStatsContactSubscribeIngestion GuideRate Limits

Overview

The Eigenvector Radar API provides programmatic access to the Zone III research intelligence platform. It is designed for research teams, enterprise architects, and automated pipelines that need to query, submit, or integrate with the paper corpus.

Base URL
text
https://multi-step-agents.com/api
Format
JSON (REST)
Auth
API Key (write)
Version
v1.0

All Endpoints

GET
/api/papers

List all papers (with optional search, chapter filter, limit)

GET
/api/papers/:id

Get a single paper by ID

POST
/api/papers

Submit a new paper (requires X-API-Key header)

GET
/api/stats

Platform statistics (paper count, sources, concepts)

POST
/api/contact

Submit a contact form message

POST
/api/subscribe

Subscribe to Zone III Intelligence newsletter

GET
/api/health

Health check endpoint

Authentication

Read endpoints (GET /api/papers, GET /api/stats) are public and require no authentication.

Write endpoints (POST /api/papers) require an API key passed in the X-API-Key header. Contact us to request API access.

bash
curl -X POST https://multi-step-agents.com/api/papers \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-api-key-here" \
  -d '{ "id": "p200", "title": "...", ... }'

Papers

GET/api/papers

Returns all papers, sorted by long-horizon relevance score. Supports optional query parameters.

Query Parameters
ParameterTypeDescription
qstringFull-text search across title, abstract, authors, topics
chapternumberFilter by chapter (1–7)
limitnumberMaximum number of results to return
bash
# All papers
curl https://multi-step-agents.com/api/papers

# Search
curl "https://multi-step-agents.com/api/papers?q=semantic+drift"

# By chapter
curl "https://multi-step-agents.com/api/papers?chapter=2"

# With limit
curl "https://multi-step-agents.com/api/papers?limit=10"
Response
json
{
  "success": true,
  "count": 300,
  "papers": [
    {
      "id": "p001",
      "title": "Lost in the Middle: How Language Models Use Long Contexts",
      "authors": ["Nelson F. Liu", "Kevin Lin", "John Hewitt"],
      "year": 2023,
      "institution": "Stanford University",
      "url": "https://arxiv.org/abs/2307.03172",
      "abstract": "...",
      "chapter": 2,
      "longHorizonScore": 0.91,
      "enterpriseScore": 0.87,
      "topics": ["context window", "long-horizon", "memory"],
      "eigenvectorCommentary": "..."
    }
  ]
}
GET/api/papers/:id

Returns a single paper by its ID, including full text and Eigenvector commentary.

bash
curl https://multi-step-agents.com/api/papers/p001
POST/api/papersRequires API Key

Submit a new paper to the corpus. The paper will be stored in the database and appear in the library after review.

Required Fields
FieldTypeDescription
idstringUnique paper ID (e.g. "p201")
titlestringFull paper title
authorsstring[]Array of author names
yearnumberPublication year
bash
curl -X POST https://multi-step-agents.com/api/papers \
  -H "Content-Type: application/json" \
  -H "X-API-Key: your-api-key-here" \
  -d '{
    "id": "p201",
    "title": "Reliable Long-Horizon Agent Execution via Hierarchical Planning",
    "authors": ["Jane Smith", "John Doe"],
    "year": 2024,
    "institution": "MIT CSAIL",
    "url": "https://arxiv.org/abs/2401.XXXXX",
    "abstract": "We present a framework for...",
    "chapter": 2,
    "longHorizonScore": 0.88,
    "enterpriseScore": 0.82,
    "topics": ["planning", "long-horizon", "hierarchical"],
    "eigenvectorCommentary": "This paper directly addresses..."
  }'

Stats

GET/api/stats
bash
curl https://multi-step-agents.com/api/stats
json
{
  "success": true,
  "stats": {
    "totalPapers": 300,
    "sources": 26,
    "concepts": 847,
    "frameworks": 12
  }
}

Contact Form

POST/api/contact

Submit a contact form message. The message will be sent to the Eigenvector Research team.

bash
curl -X POST https://multi-step-agents.com/api/contact \
  -H "Content-Type: application/json" \
  -d '{
    "name": "Jane Smith",
    "email": "[email protected]",
    "organisation": "Acme Corp",
    "subject": "Zone III Readiness Assessment",
    "message": "We are evaluating Zone III deployment..."
  }'

Newsletter Subscription

POST/api/subscribe
bash
curl -X POST https://multi-step-agents.com/api/subscribe \
  -H "Content-Type: application/json" \
  -d '{ "email": "[email protected]", "name": "Jane Smith" }'

How to Add Papers from External Sources

The Eigenvector Radar is designed to accept papers from external research pipelines. Here is a step-by-step guide to ingesting papers from arXiv, HuggingFace, ResearchGate, or any other source.

1

Request API Access

Contact us at [email protected] to request an API key for write access.

2

Prepare Paper Metadata

Structure your paper data according to the schema. The minimum required fields are id, title, authors, and year.

json
{
  "id": "p201",                          // Unique ID (required)
  "title": "Paper Title",                // (required)
  "authors": ["Author 1", "Author 2"],   // (required)
  "year": 2024,                          // (required)
  "institution": "MIT",                  // Recommended
  "url": "https://arxiv.org/abs/...",    // Recommended
  "abstract": "Full abstract text...",   // Recommended
  "chapter": 2,                          // 1-7 (see chapter guide below)
  "longHorizonScore": 0.85,              // 0.0-1.0 relevance to Zone III
  "enterpriseScore": 0.80,               // 0.0-1.0 enterprise relevance
  "topics": ["planning", "memory"],      // Array of topic strings
  "concepts": ["semantic drift"],        // Key concepts
  "frameworkMappings": ["PASF"],         // Related Eigenvector frameworks
  "eigenvectorCommentary": "...",        // Optional: your commentary
  "storageUrl": "https://..."            // Optional: PDF download URL
}
3

Chapter Assignment Guide

Ch.1
Enterprise AI Reliability Crisis— Failure modes, reliability science, production incidents
Ch.2
Long-Horizon Execution & Memory— Context windows, memory architectures, state management
Ch.3
Runtime Governance & Semantic Integrity— Policy engines, drift detection, audit trails
Ch.4
Multi-Agent Orchestration— Coordination, trust, debate protocols
Ch.5
Inference-Time Feedback & Improvement— RLHF, process rewards, self-correction, fine-tuning
Ch.6
Path to Zone III— Maturity models, enterprise readiness, deployment patterns
Ch.7
Research Synthesis— Cross-cutting themes, meta-analysis, future directions
4

Batch Ingestion (Python Example)

python
import requests
import json

API_BASE = "https://multi-step-agents.com/api"
API_KEY = "your-api-key-here"

papers = [
    {
        "id": "p201",
        "title": "Reliable Long-Horizon Agent Execution",
        "authors": ["Jane Smith"],
        "year": 2024,
        "url": "https://arxiv.org/abs/2401.XXXXX",
        "chapter": 2,
        "longHorizonScore": 0.88,
    },
    # ... more papers
]

for paper in papers:
    response = requests.post(
        f"{API_BASE}/papers",
        headers={
            "Content-Type": "application/json",
            "X-API-Key": API_KEY,
        },
        json=paper,
    )
    result = response.json()
    if result["success"]:
        print(f"✓ Ingested: {paper['title']}")
    else:
        print(f"✗ Failed: {result['error']}")

Rate Limits & Guidelines

• Read endpoints (GET): No rate limit for reasonable use. Please cache responses where possible.

• Write endpoints (POST /api/papers): Maximum 100 papers per hour per API key.

• Contact form: Maximum 10 submissions per hour per IP address.

• All responses include a success boolean and either data or error fields.

• For bulk ingestion of 500+ papers, contact us directly for a dedicated ingestion session.