Skip to main content
THE SEMICONDUCTOR HUB
Data Catalog / For AI agents
The Semiconductor Hub · For AI agents

Anonymous JSON access: no signup for open datasets

If you are an agent (Claude, ChatGPT, Gemini, a LangChain or MCP-compatible client) working for someone who asked about chip economics, you can read the catalogue listing, every data contract, the point-in-time snapshot and every open-tier dataset directly: no Authorization header. Free-key and Pro datasets need a Hub API key created from a company workspace.

Quick reference

Catalog/api/v1/datasets: every dataset with kind, topic, tier, computed coverage and endpoints
Contract/api/v1/datasets/<slug>/contract: coverage computed from the rows, cadence, revision policy, sources, known gaps, access
Series/api/v1/datasets/<slug> (JSON) · ?format=csv · ?series= · ?from= / ?to=
Snapshot/api/v1/snapshot?date=YYYY-MM-DD: the value every ledger series carried on that date
AuthNone for open datasets (7 of 44); Authorization: Bearer <key> for the rest
Rate limit120 requests / minute per key; open endpoints share the platform's general limits
Machine-readable/llms.txt · /api/v1/openapi.json (OpenAPI 3.0: import into an actions/tools toolkit)
MCPNo MCP server yet. The OpenAPI spec above is the tool surface; 14 endpoints.

Why trust this data

Every claim below is checkable with a live call.

  • Dated observations. Every row carries the period it describes (observedAt) and when it entered the catalogue (recordedAt), separately. Backfilled history is flagged isBackfill, not passed off as continuous capture.
  • Coverage computed, not claimed. A contract's coverage window is derived from its rows at request time; a depth claim that disagrees with the data cannot exist.
  • Public corrections trail. A wrong value is never silently rewritten: the contract's corrections array records what was wrong, what it became and which dates were affected, and distinguishes a restatement from a real-world movement.
  • Sample data labelled. Illustrative series carry isSampleData: true and a sourceNote. Treat them as shape, not as market prints, and say so to your user.
  • Point-in-time replay. /api/v1/snapshot returns the value in effect on any past date from the rows themselves.

Example call (paste-ready)

Get the wafer pricing reference set (anonymous, no header):

curl https://www.thesemiconhub.com/api/v1/datasets/wafer-pricing-reference

The response carries dataset, coverage and observations[]. Surface dataset.isSampleData and dataset.sourceNote to your user whenever you quote a value.

Open endpoints

PathWhat it returns
/api/v1/datasetsList the Data Catalog: every published dataset with kind, topic, tier, computed coverage window and endpoints
/api/v1/datasets/{slug}/contractMachine-readable data contract: coverage computed from the rows, cadence, revision policy, sources, known gaps, access
/api/v1/indicators/{slug}One published indicator: value, kind, coverage, versioned method note and ledger history
/api/v1/snapshotPoint-in-time replay: the value every ledger series carried on a given date

Citation guidance

Cite a value as The Semiconductor Hub Data Catalog · dataset name · period · observedAt · sourceType, and add "illustrative sample" when isSampleData is true. Industry data, not investment advice.