From fluent
Formats learner answer feedback with corrections, severity tags, and scores out of 10. Used after every practice session turn in the Fluent system.
How this skill is triggered — by the user, by Claude, or both
Slash command
/fluent:fluent-feedback-formatterThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Every practice session ends each turn with immediate feedback. Consistency matters — the learner builds mental models from the structure, and error patterns we mine from session files depend on predictable markers (❌, ✅, severity emoji). This skill defines the single feedback shape used across all Fluent practice skills.
Every practice session ends each turn with immediate feedback. Consistency matters — the learner builds mental models from the structure, and error patterns we mine from session files depend on predictable markers (❌, ✅, severity emoji). This skill defines the single feedback shape used across all Fluent practice skills.
Load this skill whenever the tutor:
fluent-learn, fluent-vocab, fluent-writing, fluent-speaking, fluent-reading, fluent-review).mistakes-db.json.Skip this skill for non-feedback output (greetings, summaries, progress reports).
{✅ or ❌} {one-line encouragement or gentle correction}
**Corrections:**
- ❌ "{wrong_part}" → **"{correct_part}"** ({category} — {brief_why})
- ✅ "{correct_part}" — {specific_praise}
**Correct version:**
"{full_correct_sentence}"
**Score: {X}/10** {emoji} {short_comment}
---
Skip the ❌ block if the answer is fully correct. Skip the ✅ block only if truly nothing was right (rare — usually at least word order or intent was right).
| Symbol | Severity | Meaning | Example |
|---|---|---|---|
| 🔴 | Critical | Breaks communication or exam-blocker | Formal/informal mix in formal email; wrong subordinate-clause word order |
| 🟡 | Moderate | Noticeable but understandable | Preposition error, missing article |
| 🟢 | Minor | Low priority | Spelling, punctuation, accent marks |
A single answer may contain multiple errors of different severity — tag each.
These feed mistakes-db.json:
grammar — word order, conjugation, clause structureformal_informal — u/je, uw/jouw, register mismatchvocabulary — wrong word, English mixing, register-wrong synonymspelling — minorprepositions — om/op/in/bij/naar/etc.articles — de/het, definite/indefinitemissing — omitted greeting, closing, required word"Close! Let's tune one word."), not a bare Wrong.."Ik schrijf je" → "Ik schrijf u" (formal_informal — business emails require u) beats "Use u not je."."This is the omdat word-order rule: verb goes last."."You didn't miss this last time — well done." when mistakes-db shows improvement.use_emojis: true by default. Keep them.After scoring, feed the score into the SM-2 update via the fluent-sm2-calculator skill: quality = floor(score / 2).
See .claude/references/feedback-template.md for fully-rendered examples (mostly-correct answer with a minor slip; critical error with severity tagging). The reference file is the authoritative version — keep it and this skill in sync if updating.
Quick pattern:
Structured, consistent feedback:
PRACTICE.md analysis + /results mining work.mistakes-db.json categories cleanly — which feeds spaced repetition, which drives the whole system.npx claudepluginhub m98/fluent --plugin fluentRuns interactive writing practice sessions with systematic error analysis across grammar, register, vocabulary, structure, and spelling. Selects scenarios matched to learner mastery level and CEFR stage.
Builds progressive writing practice with baseline assessment, CEFR level mapping, process-writing cycles, and focused corrective feedback targeting accuracy, fluency, and register control.
Analyses teacher or peer feedback against Hattie & Timperley's four-level model and rewrites it for improved specificity, actionability, and impact on student learning.