ISHOB
Machine access
ISHOB · Machine-readable knowledge archive

Knowledge prepared
for AI.

A growing machine-readable archive of structured knowledge, organized for automated discovery, retrieval, analysis, and AI applications.

Explore the Archive API Documentation

Built for machines. Understandable by people.

01 — The Archive

A growing machine-readable knowledge resource

ISHOB continuously organizes and prepares structured knowledge for automated systems. Figures are read live from the archive index.

66,034
Resources
published, API-accessible records
46,077
Structured documents
full text in sections
16
Knowledge topics
curated collections
2026-10-07 09:56 UTC
Last updated
continuously expanding

Every record: JSON · stable identifier · versioned · license & rights metadata included

02 — Tools for AI agents

9 tools, one endpoint

ISHOB runs a remote Model Context Protocol server at api.ishob.com/mcp. Assistants and agents connect once and get 3 free discovery tools plus paid content tools. Paid tools settle per call with x402 inside the MCP request: no account, no API key, no subscription.

Searchfree

search_archive

Search the ISHOB archive (titles, keywords, authors). All words must match.

{"query": "laser calibration", "limit": 10}
Discoverfree

list_topics

List knowledge topics with live resource counts.

{}
Browsefree

get_catalog_page

Browse the catalog, 100 resources per page, optionally by topic.

{"page": 0, "topic": "laser-calibration"}
Record

get_document

Record metadata, abstract, license and attribution for one resource.

{"id": "<resource id>"}
Full text

get_fulltext

Structured full text (sections) of one resource, with license.

{"id": "<resource id>"}
References

get_references

Parsed reference list of one resource.

{"id": "<resource id>"}
Web to Markdown · new

extract_webpage

Fetch a public web page live (robots.txt respected) and return clean Markdown. Not stored or redistributed.

{"url": "https://example.org/article"}
Spec search · new

search_specs

Full-text search over abstracts and full texts, with extracted technical specifications (values, ranges, tolerances, units). Numeric constraints like "50 Nm" are matched.

{"query": "torque sensor 50 Nm"}
Spec table · new

query_specs

Structured specification table filtered by quantity (e.g. accuracy, wavelength, range), unit family, unit and value range, with source document and license.

{"quantity": "wavelength", "unit": "nm", "min": 500, "max": 700}
Connect

Add ISHOB to an MCP client

Point any Streamable-HTTP MCP client at the endpoint. Free tools work immediately. For paid tools, use an x402-capable MCP client (for example the open-source @x402/mcp package): it reads the payment requirement returned by the tool, signs a USDC payment and retries the call automatically.

Server card: /.well-known/mcp.json. The same capabilities over plain HTTPS are listed in the API section.

MCP client configurationjson
{
  "mcpServers": {
    "ishob": {
      "type": "http",
      "url": "https://api.ishob.com/mcp"
    }
  }
}
New · technical specifications

The spec tools read the structured full texts and pull out measurable facts: values, ranges (“4–20 mA”), tolerances (“±0.1 %”) and their units, normalised to SI and labelled with the quantity they describe (accuracy, range, wavelength, torque…). Agents can ask for “torque sensor 50 Nm” or for every wavelength between 500 and 700 nm, and each answer cites its source record and license.

03 — How it works

From open sources to agent-ready answers

ISHOB is an automated pipeline, not a static dump. These are the stages every resource passes through before an agent can retrieve it.

  1. CollectOpen-access sources are harvested around the clock by a resumable pipeline that survives crashes and restarts without losing or duplicating work.
  2. Verify rightsEvery item is classified before publication. Reusable licenses are published in full; restricted items are kept as metadata only; unclear rights are held back.
  3. StructureContent is normalised into versioned JSON: metadata, abstract, full text in sections and parsed references, each with license, attribution and provenance.
  4. DeduplicateRecords are matched across sources by stable identifiers and content fingerprints, so each resource appears once with one permanent id.
  5. IndexA sharded search index, a paginated catalog and topic collections are rebuilt continuously and served straight from object storage.
  6. ExtractA dedicated engine indexes full texts and extracts technical specifications into a queryable table.
  7. ServeEverything is exposed over HTTPS and MCP at the edge. Discovery is free; content is paid per call with x402 in USDC.
No accounts

No sign-up, no API keys. A wallet-capable client is all an agent needs for paid calls.

Pay per call

Fractions of a cent per request. Missing resources return 404 before any payment is asked; failed deliveries are never charged.

Rights first

Licenses travel with every record. Live web extraction honours robots.txt and is never stored or redistributed.

Discoverable

MCP server card, OpenAPI, llms.txt, ai-data.json, sitemaps and the x402 Bazaar describe every capability.

04 — Built for machines

Built for machines

ISHOB is designed to be discovered and accessed programmatically. AI agents and automated systems can explore available topics, inspect structured metadata, search the archive, and request available resources through machine-readable interfaces.

Machine discovery

robots.txt, sitemaps, /.well-known/ai-data.json, /.well-known/mcp.json, llms.txt and OpenAPI describe the archive.

Programmatic access

Plain HTTPS or MCP. No accounts, no API keys, no browser workflow.

Structured formats

Consistent JSON records with stable identifiers, sections, references and rights fields.

API access

Catalog, topics and search are free; eligible content is served per request.

GET/apiArchive description, endpoints, formats, access modelfree
GET/api/catalogPaginated catalog of all resources (100 per page)free
GET/api/topicsKnowledge topics with live countsfree
GET/api/searchSearch titles, keywords and authors; filter by topic, year, licensefree
GET/.well-known/ai-data.jsonMachine discovery documentfree
GET/openapi.jsonOpenAPI 3.1 descriptionfree
MCP/mcpRemote MCP server (Streamable HTTP) — free discovery tools, x402-paid content toolsmixed
GET/api/document/{id}Record metadata, abstract and rights
GET/api/document/{id}/fulltextStructured full text in sections
GET/api/document/{id}/referencesParsed references
POST/v1/extractLive fetch of a public page as Markdown (robots.txt respected, not stored)
POST/v1/searchFull-text search with technical-spec matching, e.g. 'torque sensor 50 Nm'
POST/v1/specsStructured specification table: quantity, unit, value range, source and license
05 — Explore

Explore the archive

Resources are organized into knowledge topics. Each topic is a machine-readable collection with its own catalog and search scope.

06 — Prepared knowledge

Structured knowledge, prepared for machines.

ISHOB organizes knowledge into consistent, machine-readable records designed for automated discovery, retrieval, analysis, and AI workflows.

Full records add the abstract, structured sections, references, and complete license, attribution and provenance fields. Data & rights.

Example recordapplication/json
// EXAMPLE RECORD — illustrative values, real field structure
{
  "id": "doi-1a2b3c4d5e6f7a8b9c0d",
  "title": "Example resource",
  "year": 2025,
  "topics": ["example-topic"],
  "keywords": ["measurement", "uncertainty"],
  "license": "CC-BY-4.0",
  "has_fulltext": true,
  "has_references": true,
  "access": "paid",
  "content_endpoint": "/api/document/doi-1a2b3c4d5e6f7a8b9c0d",
  "fulltext_endpoint": "/api/document/doi-1a2b3c4d5e6f7a8b9c0d/fulltext"
}
07 — Workflows

Ready for AI workflows

Structured resources are organized for automated discovery and programmatic access. Explore the archive, inspect available metadata, and retrieve resources through the ISHOB API, or run the same steps as MCP tool calls (list_topics → search_archive → get_fulltext).

  1. Discover/.well-known/ai-data.json
  2. Explore/api/topics
  3. Search/api/search?q=…
  4. Request/api/document/{id}
  5. Retrieve/api/document/{id}/fulltext
08 — Access

Machine-to-Machine Access

Discovery is free. Eligible resources are delivered per request using the x402 HTTP payment protocol: compatible clients receive a machine-readable payment requirement and complete it programmatically.

Over MCP the same exchange happens inside the tool call: the tool returns the payment terms, the client signs, and the result arrives with a settlement receipt. Network: Base · settled in USDC. Missing resources return 404 before any payment is requested; failed deliveries are not charged. Clients without x402 support can still use all free endpoints.

  1. AI AGENTAny HTTP or MCP client
  2. DISCOVERTopics, catalog, search — free
  3. REQUEST RESOURCEGET /api/document/{id}/fulltext
  4. PAYMENT REQUIREDHTTP 402 with PAYMENT-REQUIRED terms
  5. PAYMENT VERIFIEDSigned payment verified and settled
  6. RESOURCE DELIVERED200 OK · structured JSON
Metadata
$0.001
Full text
$0.01
References
$0.002
Topic bundle
$0.5
Web extract
$0.0005
Spec search
$0.001
About

ISHOB — Intelligent Structured Human & Open Knowledge Base

ISHOB is a continuously expanding archive of structured knowledge prepared for machine discovery and AI consumption. Resources are published only when their reuse rights are verified, and each record keeps its license, attribution and provenance metadata.

The figures on this page are the archive's actual current contents. Discovery data is free; some resources are delivered through paid machine access.

Start here: /api → /api/catalog → /api/topics → /api/search