FEED

Nutrition research · public warehouse

Unclassified

01 collect—
02 normalize—
03 preserve—
04 search

Records

—

Sources

—

New · 24h

—

Live

Pulse

Same warehouse · same provenance

Pulse

The data, right now.

Loading live data.

FEED

Public-data collection without the platform lock-in.

FEED is an extensible framework for collecting, normalizing, preserving, and searching public information. Connect authorized APIs, open datasets, public websites, research sources, and other approved data feeds through a common source-adapter interface.

Every record keeps its source, timestamp, and provenance so analysts and researchers can understand where the data came from and how it changed over time.

How it works · Open the desk · View live data · Request API access

Sources

One interface. Many sources.

Different sources require different acquisition methods. FEED treats them as interchangeable adapters. Each adapter can carry its own authentication, rate limits, retention requirements, and usage policies.

Official APIs
Research APIs
Open government datasets
RSS and syndication feeds
Public websites where automated collection is permitted
Bulk open-data archives
Customer-authorized commercial feeds

Evidence

Preserve the evidence.

Raw evidence

The source response or observation as it was collected.

Normalized data

A common schema that makes information searchable across sources.

Derived intelligence

Topics, entities, relationships, trends, and other enrichment, kept connected to the underlying source.

Live data

Public nutrition research.

FEED is collecting public posts, news, stills, recalls, papers, and open datasets on food, nutrition, eating, hunger, and related research. Every row keeps its source, credit, and time. The same adapters run other authorized public-data missions.

Pulse

The data, right now.

Loading live data.

Search the data.

Loading live data.

Sources

One interface. Many sources.

Loading live data.

Schedule

It keeps collecting.

Loading live data.

Connect

Built for the model in your pipeline.

REST, CSV, JSON Schema, a RAG pack, and a hosted MCP server. One warehouse behind all of them.

REST

Search, ask, stats, signals, sentiment, sources, runs, QC, and CSV or JSON export. The same warehouse this page reads. OpenAPI and JSON Schema ship with the key.

curl -H "Authorization: Bearer $FEED_KEY" \
  "https://deadindex.com/api/feed/records?q=school%20lunch&topic=policy&limit=20"
curl -H "Authorization: Bearer $FEED_KEY" \
  "https://deadindex.com/api/feed/export.csv?topic=safety&limit=2000" -o recalls.csv
curl -H "Authorization: Bearer $FEED_KEY" -X POST https://deadindex.com/api/feed/ask \
  -d '{"question":"sentiment on SNAP this month"}'
curl -H "Authorization: Bearer $FEED_KEY" https://deadindex.com/api/feed/sentiment
curl -H "Authorization: Bearer $FEED_KEY" -X POST https://deadindex.com/api/feed/sentiment \
  -d '{"text":"SNAP cuts mean we skip dinner two nights a week now."}'

MCP

A hosted, read-only MCP server over the same warehouse. Tools: search, ask, stats, signals, sources, QC. Add it to Cursor, Claude, or any MCP client.

{
  "mcpServers": {
    "feed": {
      "url": "https://deadindex.com/api/feed/mcp",
      "headers": { "Authorization": "Bearer <your FEED key>" }
    }
  }
}

LLM

A RAG pack: one chunk per matching row with source, URL, date, facets, and sentiment, plus a tool catalog. The Ask view on the desk runs the same path an agent would.

Reads on this page are open. Pipelines use a key; request one below. Write access stays with the collectors.

API

Request FEED access.

A key is for a pipeline you run. Tell us how you would use it. We review each request.

Access

Talk to us about FEED.

Built to plug in, not take over.

Public-data collection without the platform lock-in.

Every record keeps its source, timestamp, and provenance so analysts and researchers can understand where the data came from and how it changed over time.

One interface. Many sources.

Official APIs

Research APIs

Open government datasets

RSS and syndication feeds

Public websites where automated collection is permitted

Bulk open-data archives

Customer-authorized commercial feeds

Built for the model in your pipeline.

REST, CSV, JSON Schema, a RAG pack, and a hosted MCP server. One warehouse behind all of them.

REST

Search, ask, stats, signals, sentiment, sources, runs, QC, and CSV or JSON export. The same warehouse this page reads. OpenAPI and JSON Schema ship with the key.

curl -H "Authorization: Bearer $FEED_KEY" \
  "https://deadindex.com/api/feed/records?q=school%20lunch&topic=policy&limit=20"
curl -H "Authorization: Bearer $FEED_KEY" \
  "https://deadindex.com/api/feed/export.csv?topic=safety&limit=2000" -o recalls.csv
curl -H "Authorization: Bearer $FEED_KEY" -X POST https://deadindex.com/api/feed/ask \
  -d '{"question":"sentiment on SNAP this month"}'
curl -H "Authorization: Bearer $FEED_KEY" https://deadindex.com/api/feed/sentiment
curl -H "Authorization: Bearer $FEED_KEY" -X POST https://deadindex.com/api/feed/sentiment \
  -d '{"text":"SNAP cuts mean we skip dinner two nights a week now."}'

MCP

A hosted, read-only MCP server over the same warehouse. Tools: search, ask, stats, signals, sources, QC. Add it to Cursor, Claude, or any MCP client.

{
  "mcpServers": {
    "feed": {
      "url": "https://deadindex.com/api/feed/mcp",
      "headers": { "Authorization": "Bearer <your FEED key>" }
    }
  }
}

LLM

A RAG pack: one chunk per matching row with source, URL, date, facets, and sentiment, plus a tool catalog. The Ask view on the desk runs the same path an agent would.

Reads on this page are open. Pipelines use a key; request one below. Write access stays with the collectors.