voice-refine

Transform verbose voice input into optimized Claude prompts

Install with Tessl CLI

npx tessl i github:FlorianBruniaux/claude-code-ultimate-guide --skill voice-refine

What are skills?

1.21x

Quality

58%

Does it follow best practices?

Impact

91%

1.21x

Average score across 3 eval scenarios

Optimize this skill with Tessl

npx tessl skill review --optimize ./examples/skills/voice-refine/SKILL.md

SKILL.md

Review

Evals

Evaluation results

93%

50%

Structuring a Voice-Dictated Feature Request

Output format and information preservation

Criteria

Without context

With context

Contexte section

100%

Objectif section

100%

Contraintes section

100%

Output attendu section

100%

Single-sentence Objectif

100%

Stack in Contexte

100%

Redux context preserved

100%

Background notification edge case

100%

Per-category toggle preserved

100%

Filler removed

100%

Significant compression

Without context: $0.1795 · 49s · 11 turns · 16 in / 2,670 out tokens

With context: $0.5693 · 2m · 25 turns · 2,495 in / 7,198 out tokens

90%

Processing a Performance Debugging Voice Note

Compression and filler removal

Criteria

Without context

With context

Filler words absent

100%

Hedging phrases absent

100%

Politeness padding absent

100%

Performance symptom preserved

100%

Tech stack preserved

100%

Key tables preserved

100%

Query volume issue preserved

100%

Caching solution preserved

100%

Indexing issue preserved

100%

Target compression achieved

Without context: $0.5576 · 2m 20s · 41 turns · 317 in / 6,853 out tokens

With context: $0.4797 · 1m 39s · 21 turns · 2,671 in / 5,949 out tokens

92%

Refining a Spanish Voice Dictation for an International Engineering Team

Flag behavior: --verbose and --en

Criteria

Without context

With context

English output

100%

No Spanish text

100%

Verbose detail: data scale

100%

Response time constraint

100%

Recommendation count preserved

100%

Both algorithm types

100%

A/B testing preserved

100%

Full tech stack present

100%

Model update requirement

100%

Less compression than standard

100%

Standard sections present

Without context: $0.4627 · 2m 2s · 33 turns · 65 in / 6,435 out tokens

With context: $0.4591 · 1m 38s · 21 turns · 2,491 in / 5,445 out tokens

Evaluated: 5 days ago
Agent: Claude Code
Model: Claude Sonnet 4.6

Table of Contents

Structuring a Voice-Dictated Feature Request Processing a Performance Debugging Voice Note Refining a Spanish Voice Dictation for an International Engineering Team

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.