INFRASTRUCTURE

One graph. Three surfaces.
Infrastructure for skills AI.

Skillberg exposes 158,000 skills through a REST API, inside Claude and ChatGPT via MCP, and as managed access to the Neo4j graph. Hosted in France. AI Act compliant by design.

See the quickstart Connect Claude or ChatGPT

● api · ~99.9%region: fr-parp95 ~220msopenapi 3.1v0.4

3 surfaces · 1 graph

01 · Surfaces

Three surfaces, one graph.

One backend, three ways to use it, depending on your technical context.

REST

REST API

HTTPS endpoints for skill search, enrichment and matching. OpenAPI 3.1, x-api-key, rate limits per tier.

# Skill search
curl https://api.skillberg.app/v1/skills/search?q=python \
  -H "x-api-key: sk_live_…"

Search · Enrichment · CV ↔ job matching
Free Discovery tier as soon as you sign up
EU regions (fr-par by default)

API documentation

MCP

MCP connector for Claude and ChatGPT

Compatible with Claude, ChatGPT and any MCP client. Your agents query the graph in natural language, with OAuth and per-user traceability.

// Claude prompt
"Find the skills related to the senior Data Scientist job."

15 MCP tools (search_skills, skills_gap, career_transitions, …)
OAuth 2.1 + DCR + PKCE
Usage tracked per API key

See the connector

GRAPH

Managed graph

Read-only Cypher / Neo4j access for Omni customers. Per-customer replication, ready-to-use embeddings, monthly updates.

// Cypher · transitive prerequisites
MATCH (s:Skill)-[:PREREQUISITE_OF*1..3]->
  (t:Skill {label:'Deep Learning'})
RETURN s

8 frameworks covered (ESCO, ROME, O*NET, COMPETENT, SSF, CO-SISPE, AI4ESCO, PT)
18,073 validated cross-standard junctions
1536-dim embeddings available

Talk to an expert

02 · Architecture

How it fits together.

One chain, from the Neo4j graph to your application.

Skillberg API

handles authentication, routing and usage-based billing.

Graph engine

runs graph queries, CV enrichment and match scoring.

Graph

hosted on Scaleway (Paris), replicated for availability.

03 · The proprietary graph

Eight global frameworks,
joined by 18,073 validated bridges.

Built over 24 months of doctoral R&D (CIFRE), updated monthly by our engineering team. Every junction is traceable, dated and reviewable.

ESCOROMEO*NETCOMPETENTSingapore SSF

Dive into the graph

KNOWLEDGE GRAPH · v0.4LIVE

canonical skills

occupations

frameworks

cross-standard junctions

04 · Quickstart

Get started in 15 minutes.

One endpoint, one key, your first integration.

GET /v1/skills/search

curl https://api.skillberg.app/v1/skills/search?q=python \
  -H "x-api-key: sk_live_..."

200 · application/jsonschema

{
  "query": "python",
  "took_ms": 84,
  "skills": [
    {
      "id": "sb:skill/programming/python",
      "label": "Python",
      "score": 0.97,
      "referentials": {
        "esco":  "S6.1.1.1",
        "rome":  "M1805",
        "onet":  "15-1252.00"
      },
      "junctions": 47,
      "prerequisites": ["sb:skill/cs/algorithms"]
    },
    …
  ]
}

See the full response

Create a free API key Read the OpenAPI spec

05 · Sovereignty

European sovereignty,
compliance by design.

Hosted in France

Scaleway, fr-par region. No transfers outside the EU.

fr-par · ISO 27001

European models

Mistral LLMs for enrichment and scoring.

Mistral · self-hosted

AI Act ready

Explainability through graph traceability. Ready for Annex III (deadline December 2, 2027).

AI Act · Annex III

GDPR compliant · ISO 27001 in progress · Audits on request.

06 · Partners & recognition

CDE

BPI France

Euratechnologies

Scaleway

Mistral AI

07 · Get started

One line of code
is all it takes.

Free console Connect Claude or ChatGPT

Need Omni access to the graph? contact@skillberg.app

One graph. Three surfaces.Infrastructure for skills AI.