This directory holds the detailed description of the designed behaviour of CommentRake. A capability counts as implemented only when the documentation says so explicitly or the code and the tests confirm it.
- Product concept — purpose, boundaries, status and the overall order of work.
- Project map — the shared hierarchical model, the metrics and the views in the different interfaces.
- Comment analysis — file discovery, parsing, kinds of comments, language adapters and classification rules.
- Supported languages — a reference: how a file is given a language, which languages are parsed, where the grammars come from.
- Rule reference — what each rule looks for, how strong its finding is, how to turn it on and off, and the review-priority indicators.
- Command-line reference — every command, its options, the exit codes and what each one writes.
- Protected constructs by language — the directives, documentation and structural markers of the first wave of languages that ordinary rules must not remove.
- Safe file editing — result verification, the EditPlan, dry-run, apply, archive and restore.
- Reviewing changes before a commit — the scope by changed files and lines, interaction with the index, the behaviour of the pluggable check and the performance requirements.
- Threat model for the modifying operations — the assets, the untrusted conditions and the measures for archive, apply and restore.
- Project history — SQLite, scan snapshots, comparison of states and comments that came back.
- Cleanup statistics — what is measured, the list of counting methodologies and why older numbers stay explainable.
- Integration with coding agents — preventing new noise, the machine interface and the architectural translation hook.
- AI guidance package — English, platform-neutral instructions for the user's own AI agents.
- Interface design — the shared navigation, the mandatory mockups and the implementation boundaries of the GUI, TUI and CLI.
- Configuration — the TOML schema, exclusions, language mappings, tests, rules and how the format evolves.
- Quality assurance — test levels, regression fixtures, invariants, performance and release.
- Development plan — implementation stages, mandatory checks and readiness criteria.
- Development environment — Python, Conda, the approved dependency groups and reproducible environment creation.
- Architecture — the layers, the dependencies and the technical boundaries of the system.
- Project home page (Russian version — README.ru.md)
- Contribution guidelines
TASKS.internal.md— the private working plan; it is not part of the public documentation and must not be published.