what is ravel fr?

Ravel is a command line tool, written in Go, that builds a local knowledge graph out of a codebase so both developers and AI coding assistants can understand a project without reading every file from scratch. It ships as a single binary with no Python runtime needed, and it runs entirely offline. After installing it, a person runs a build command inside a repository and Ravel creates a local graph that captures which files, functions, and types exist and how they connect to each other. The tool separates what it is confident about from what it is guessing. Facts pulled directly from parsing the code are labeled extracted, matches based on naming patterns are labeled inferred, and anything unresolved stays unresolved rather than being presented as fact. There is an optional feature where an AI agent can add extra context, such as a description of the system's architecture or a suggested reading tour through the code, but this enrichment is always marked separately from the parser's own output so a reader can tell which parts came from code analysis and which came from an agent's interpretation. Before Ravel reads anything, it runs an audit step that shows exactly which files it plans to look at, and it is built to automatically skip secrets, dependency folders, and files a project has chosen to ignore. Once a graph exists, a user can search it for a specific file, package, type, function, or method, or ask a natural language question and get back a compact, connected answer built from the graph rather than a fresh scan of the whole repository. Ravel also plugs into AI coding assistants such as Codex, Claude, Cursor, and several others through an installable skill bundle, so those tools can query the graph before searching a repository broadly, and check what a change might affect before it is made. It supports Go with deep analysis and many other languages through Tree-sitter grammars. It is released under the MIT license. The full README is longer than what was shown.