In this paper, the authors consider the process of building and analyzing a skills graph based on vacancies data from the job search portal. The connection between vacancies and skills is identified and formalized as labor market and employers' requirements to employee's qualification in the intellectual educational ecosystem model. The authors describe the collector program which can fetch vacancies data from job search sites with an example of site HeadHunter. Jupyter Notebook interactive notebooks and different Python-based libraries, such as Pandas and Numpy, are used for data processing. As a result, a skills graph is built and stored in the neo4j graph database platform. With the help of the Gephi application, the next properties of this graph are determined: degree distribution, the eigenvector centrality, closeness centrality, betweenness centrality, PageRank coefficient, and modularity. The graph is drawn using the ForceAtlas2 layout algorithm. The modularity coefficient for the vertices is used to find the main clusters of skills. The number of such skills clusters is almost the same as the count of specialties on the example job search portal. It results in the ability to determine specialty skillsets and interspecialty skills using vacancies data from job search sites.