The article introduces jsongrep, a tool for querying JSON data efficiently. Built in Rust, it utilizes a deterministic finite automaton (DFA) to process queries, allowing for fast searches through JSON documents without the overhead of backtracking or recursion. The author emphasizes the speed of jsongrep, comparing its performance to other JSON query tools like jq, jmespath, and jsonpath-rust. Unlike these alternatives, which interpret queries and frequently revisit nodes, jsongrep compiles queries into a DFA before executing them, enabling a single-pass search with constant time complexity per input symbol. The article illustrates jsongrep's query capabilities using a sample JSON document. It explains how to construct queries with paths, wildcards, and recursive descent, providing concrete command-line examples. Users can select nested fields, match any key or index, and filter results based on specific criteria. The tool is installed via cargo or Homebrew, making it accessible across platforms. The author also delves into the internals of jsongrep, outlining its five-stage pipeline: parsing the JSON document, parsing the user's query into an abstract syntax tree (AST), constructing a nondeterministic finite automaton (NFA), determinizing it into a DFA, and finally walking the JSON tree to collect matches. This design leverages the properties of regular languages to ensure efficient query processing. The article concludes with an acknowledgment that jsongrep is still developing, lacking some of the broader functionalities found in tools like jq, but it positions itself as a specialized search tool rather than a transformation tool.