Krrkt
Check Russian text quality using the krrkt CLI tool — gives an objective 0-10 score with categorized findings. Use when: (1) user writes or edits Russian text and asks to check quality, (2) user says "проверь качество текста", "проверка качества", "стоп-слова", "krrkt", "check text quality", (3) user asks to score Russian writing style, (4) user opens a Russian .md/.txt file and asks for quality feedback, (5) user asks you to write Russian text — check it with krrkt before delivering. Requires krrkt CLI (pip install krrkt). NOT for English text, translation, or spelling/grammar (use ru-textovod for that).From its SKILL.md
npx -y skills add miolamio/agent-skills --skill krrktAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 13 stars13 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its file declares
Copied from the file, not written here
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
2.5 KB, as published. Nobody here has run it
krrkt — Russian Text Quality Checker
CLI tool that scores Russian text 0-10 and flags stop words, stamps, bureaucratese, and other informational style issues. 1200+ pattern dictionary, Glvrd-compatible scoring.
Quick check
krrkt check --no-llm "Text to check"
Always use --no-llm (instant, no API needed). For files: krrkt check --no-llm /path/to/file.txt.
JSON output: krrkt check --no-llm --format json "Text"
Score only: krrkt score --no-llm "Text"
Score: 7.5+ green, 5-7.4 orange, below 5 red
Workflow
Checking existing text: Run krrkt check --no-llm, show the score and key findings (weight 80+). Suggest fixes based on finding descriptions. krrkt already categorizes issues — use its output directly rather than re-analyzing from scratch.
Writing new Russian text: Write it, then run krrkt check --no-llm on your output. If score is below 7, fix the flagged issues and re-check. Deliver the final version with the score.
Checking .md files: Extract prose only — skip YAML frontmatter, code blocks, HTML, markdown syntax.
Batch: for f in *.md; do echo "=== $f ===" && krrkt score --no-llm "$f"; done
Interpreting findings
krrkt output is self-explanatory: each finding has a rule name, weight, and often a description with fix advice. Focus on ■ markers (weight 80-100) — those are the real problems. ▲ (1-79) are suggestions. ○ (0) are informational.
The description field in JSON output often contains the specific fix (e.g., "Замените на «общение»" for pompous words). Use it.
Install
If krrkt is not found: pip install krrkt or uvx krrkt check --no-llm "test".