EVIDENCE/PCB
For developers + AI assistants

PCB research.
Ready for your code.

Query documented capabilities and their evidence through a public JSON API or remote MCP server. Every claim keeps its source, production conditions, and review dates.

Free public previewRead-only · v1.0No account or API key requiredLoading dataset counts…
This first dataset covers Champion Asia Group. It contains manufacturer-published research, with stated limitations and expired certificate copies. Headquarters country and factory geography are separate; the source set does not establish a Taiwan factory. Text filters do not infer numeric capability or available capacity.

Build with the API

Standard HTTPS GET requests. Search companies, filter individual capability claims, and retrieve their cited sources.

https://champion-pcb-capabilities.chipmoreno.chatgpt.site/api/v1

Connect an AI assistant

A public Streamable HTTP MCP endpoint with six tools for company, capability, and source research.

https://champion-pcb-capabilities.chipmoreno.chatgpt.site/api/mcp
Connection guide

API endpoints

Base URL: /api/v1. All endpoints are public and accept GET. JSON responses use data and meta; list responses also include pagination.

Method / pathWhat it returns
GET /companiesCompany metadata; filter by query, headquarters_country, or capability_query.
GET /companies/{company}One company, its evidence counts, and API links. Use a company slug or UUID.
GET /companies/{company}/capabilitiesThat company's capability claims with conditions, citations and dates.
GET /companies/{company}/sourcesInventoried pages/PDFs and review/access status; filter by status or language.
GET /capabilitiesSearch claims across companies. Filters apply within each claim record.
GET /claims/{claim}One claim plus its source-inventory record. Use its record_id.
GET /fieldsAvailable field labels and claim counts. Optional company filter.
GET /metaAPI version, dataset counts, cache snapshot timestamp, and evidence note.

Capability filters: query, field, production_tier, material, factory, and evidence_status. The global /capabilities endpoint also accepts company (slug or UUID). Text filters are case-insensitive substrings; evidence/status and headquarters-country filters are exact, case-insensitive matches.

JavaScript

const base = "https://champion-pcb-capabilities.chipmoreno.chatgpt.site/api/v1";
const response = await fetch(
  `${base}/capabilities?company=champion-asia&query=impedance`
);
if (!response.ok) throw new Error(`HTTP ${response.status}`);
const { data, pagination } = await response.json();
console.log(data, pagination);

Python — standard library

import json
from urllib.request import Request, urlopen

base = "https://champion-pcb-capabilities.chipmoreno.chatgpt.site/api/v1"
request = Request(base + "/companies/champion-asia",
                  headers={"User-Agent": "pcb-evidence-client/1.0"})
with urlopen(request) as response:
    result = json.load(response)
print(result["data"]["name"])

Use a descriptive User-Agent for Python integrations. The hosting gateway currently rejects Python's default urllib header with HTTP 403.

Pagination and errors

limit defaults to 25 and is bounded to 1–100. offset defaults to 0. Read has_more and next_offset to request the next page. Invalid parameters return HTTP 400; missing companies/claims return 404; write methods return 405; database unavailability returns 503. Error responses contain error.code and error.message.

{ "pagination": {
  "total": 91, "limit": 25, "offset": 0,
  "has_more": true, "next_offset": 25
} }

The server caches database snapshots for 30 seconds and HTTP responses may also be cached for 30 seconds. Inspect meta.snapshot_loaded_at. The preview currently supports up to 10,000 rows per underlying table; larger datasets require a query strategy update.

Try a real request

Ready to run.
Choose an example and run it to see the live JSON response.

MCP connection guide

In a client that supports remote Streamable HTTP servers, add this server URL:

https://champion-pcb-capabilities.chipmoreno.chatgpt.site/api/mcp

No API key or login is required for this public preview. Client configuration syntax varies; use its remote HTTP server setting. The server supports MCP 2026-07-28 plus legacy initialization versions 2025-11-25, 2025-06-18 and 2025-03-26. It is stateless and returns JSON responses. GET/DELETE streams are not supported.

search_companies

Search companies and optional capability mentions, with scoped matching examples.

get_company

Get a company's metadata and evidence counts by slug or UUID.

search_capabilities

Filter documented claims by company, field, material, factory, tier, or evidence status.

get_claim_evidence

Retrieve one claim and its original source record.

get_company_sources

List inventoried pages and PDFs with review/access states.

list_capability_fields

Discover available field labels and record counts.

Modern MCP request example

curl 'https://champion-pcb-capabilities.chipmoreno.chatgpt.site/api/mcp' \
  -H 'Content-Type: application/json' \
  -H 'Accept: application/json, text/event-stream' \
  -H 'MCP-Protocol-Version: 2026-07-28' \
  -H 'Mcp-Method: tools/call' \
  -H 'Mcp-Name: search_capabilities' \
  --data '{
    "jsonrpc": "2.0", "id": 1, "method": "tools/call",
    "params": {
      "name": "search_capabilities",
      "arguments": {"company":"champion-asia","query":"impedance","limit":2},
      "_meta": {
        "io.modelcontextprotocol/protocolVersion": "2026-07-28",
        "io.modelcontextprotocol/clientInfo": {"name":"your-client","version":"1.0"},
        "io.modelcontextprotocol/clientCapabilities": {}
      }
    }
  }'

Modern clients can call server/discover or tools/list with the same per-request metadata and matching method header. Legacy clients use initialize, then notifications/initialized, tools/list, and tools/call; no session ID is needed. Browser MCP requests must originate from this site; external server and desktop clients can connect without an Origin header. The JSON API supports cross-origin browser requests.

The data contract

  • normalized_value is the extracted readable value. It is currently text; numeric units and limits are not inferred.
  • factory_or_entity, production_tier, material_or_product, and conditions_and_scope preserve the claim's context.
  • source_url, source wording, review dates, translation method and evidence status accompany the claim.
  • source_last_checked_at records source review; it does not establish when the manufacturer published the information.
  • Product cases, factory claims, expansion plans and certificate copies retain their source labels. Missing values remain unknown/unpublished.
  • Company search by a capability mention is a research starting point. Do not combine examples from different factories, materials or production tiers to assert qualification.

Browse the company sample · Source audit · Download this integration guide · MCP transport specification

Copied MCP URL.