The tokei Code Statistics Counter skill provides agents with codebase analysis capabilities for understanding project composition, tracking code growth, and generating reports. Running tokei ./src produces a breakdown showing each language detected in the directory tree with file counts and line statistics. The tool accurately distinguishes between code, comments, and blank lines using language-specific parsing rules rather than simple heuristics, handling nested comments, multi-line strings, and language-specific syntax correctly.

tokei outputs data in multiple structured formats including JSON, YAML, and CBOR (via -o json), enabling programmatic consumption by other tools and pipelines. This makes it straightforward to integrate tokei results into CI pipelines for tracking code metrics over time, generating project documentation, or feeding into dashboards. The tool supports filtering by language type (-t rust,python), sorting by various columns (files, lines, code, comments), and processing multiple directories or files in a single invocation.

Results can be saved and loaded from previous runs (-i flag), enabling diff-based analysis to see how code statistics change between commits or releases. tokei respects .gitignore files and supports custom exclusion patterns. Installation is available via cargo (cargo install tokei), Homebrew, Chocolatey, apt, and most other package managers. The tool is distributed under a permissive license, has over 14,000 GitHub stars, and millions of downloads across package managers. It provides a reliable, fast foundation for any codebase health or composition analysis task.