Disassembled binary code can be turned into a graph of instructions linked by possible execution flow (Control Flow Graph). Based on academic research on malware detection through graph matching and facing large numbers of similar files to analyze, we aim to provide accurate results to an analyst working on malware families. Our approach is a YARA-like detection tool: GRAP matches user-defined graph patterns against the CFG of a given code. GRAP is a standalone tool that takes patterns and binary files, uses a Capstone-based disassembler to obtain the CFGs from the binaries, then matches the patterns against them. Patterns are user-defined graphs with instruction conditions (“opcode is xor and arg1 is eax”) and repetition conditions (3 identical instructions, basic blocks…). The algorithm solves a simplified version of the subgraph isomorphism problem, allowing the matching to be very quick. It can be used to find generic patterns such as loops and to write signatures to detect malware variants. We also developed a plugin giving IDA the capabilities to detect and browse matches directly within the GUI. Python bindings are available to create scripts based on GRAP and extract valuable information (addresses, instructions) from matched parts. In this talk, we will introduce the algorithms used and then focus on practical use cases: detect common patterns (from the command line or within IDA), create a malware pattern, and extract information from matched instructions. The tool and the plugin will be released under an open source license.