INL/EXP-25-86503-Revision-0 Evidence-based Graph Adversary Mapping (EGRAM) August 2025 Paul-Ann Shannon Francis INL is a U.S. Department of Energy National Laboratory operated by Battelle Energy Alliance, LLC INL/EXP-25-86503-Revision-0 Evidence-based Graph Adversary Mapping (EGRAM) Paul-Ann Shannon Francis August 2025 Idaho National Laboratory Idaho Falls, Idaho 83415 http://www.inl.gov Prepared for the U.S. Department of Energy Under DOE Idaho Operations Office Contract DE-AC07-05ID14517 Evidence-based Graph Adversary Mapping (EGRAM) Intern: Paul-Ann Francis (D520) | University of South Florida | Mentors: Rita Foster (D520), Bryan Beckman (D520), Doug Buddenbohm (D330) APT41, Brass Typhoon, or BARIUM? Adversary Data in STIG • Problem statement: Cybersecurity companies such • Adversary data can be associated with malware toolsets, as CrowdStrike, Dragos, Microsoft and Unit 42 attack patterns, indicators and other context using graph categorize Advanced Persistent Threats (APTs) using structures. their own naming schemes. • These APTs are mapped to different malware sources and campaigns, all from differing sources. • Inconsistent mapping causes confusion and adds further obscurity around these groups, making it difficult to track and mitigate APT cyberattacks. • Current adversary data is stored in CSV files which Figure 2: The image (left) shows a condensed STIG of all the adversary does not show the rich relationships in the data. [1] data used. The image (right) is a snippet of adversary data displaying the relationship between threat actor groups, their attack patterns, and tools Solution used. • EGRAM remediates the mapping challenge by collecting, Adversary Data in Neo4J updating and converting adversary data and their sources into a valid, codified STIX v2.1 bundle which is • Utilizing advanced graph theoretics, STIX bundles then stored in a Neo4j graph database. representing 550+ APTs are translated into the Neo4j database running on Docker. • EGRAM exists as Python code and a Jupyter Notebook that acts as a searchable, evidence-based, source of intelligence for APT groups’ artifacts and cyber campaigns. • Utilizes graph traversal methods and centrality analysis Figure 3: Both images represent snippets of the data shown in the right image of to generate actionable information as a Structured Threat Figure 2, as it is, inside of the Neo4j database. Intelligence Graph (STIG), based on user queries. • The Neo4j database underlying STIG was put through an User Queries in EGRAM ACID test by Faith Coslett to verify capabilities. • EGRAM uses the Neo4j Python driver library to establish a connection to the Neo4j database from Python. What is an Adversary Group? • With user inputs in Python, queries are made in the • APTs are state entities or government-backed, cyber CYPHER language using the cypher-shell tool adversary groups who conduct multiple campaigns such as espionage, IP theft, and critical infrastructure Future Works sabotage. [2] • More user defined queries will be added such as attack • Conducting ransomware cyberattacks such as WannaCry patterns and IP addresses used by specific threat actors, and NotPetya, with APTs operating out of countries such malware toolsets employed by certain threat groups, and as China, Russia, North Korea and Iran. indicators linked to cyber campaigns. References [1] “APT Groups and Operations,” Google Docs. Accessed: Jul. 24, 2025. [Online]. Available: https://docs.google.com/spreadsheets/d/1H9_xaxQHpWaa4O_ www.inl.gov [2] “Nation-state threats,” Nation-State Threats | Cybersecurity and Infrastructure Security Agency CISA, https://www.cisa.gov/topics/cyber-threats-and- advisories/nation-state-cyber-actors (accessed Jul. 24, 2025). INL/EXP-25-86503