What problem does it solve? Understanding an entire codebase requires reading dozens of files across many directories, which is slow and error-prone. This Skill packs a whole repository into one structured output file so an AI or human can reason about architecture, cross-file consistency, and unfamiliar codebases in a single pass. ## Core Features & Use Cases - Repository Packing: Bundle a local directory or remote GitHub repo into XML, Markdown, JSON, or plain text output with directory structure and file summaries. - Token Compression: Use Tree-sitter-based compression to strip implementation bodies and keep only signatures, reducing token count by roughly 70% for large repos. - Configurable Filtering: Respect .gitignore and custom ignore patterns, with token counting via the o200k_base encoding to keep output within context limits. - Use Case: Before writing an architecture spec that spans multiple modules, run repomix on the project root and feed the resulting XML blob into the planning stage instead of opening files one by one. ## Quick Start Ask the AI to pack the current repository with repomix and summarize its architecture from the generated output file.
