CtrlK
BlogDocsLog inGet started
Tessl Logo

skill-creator

Guide for creating effective skills. This skill should be used when users want to create a new skill (or update an existing skill) that extends Claude's capabilities with specialized knowledge, workflows, or tool integrations.

Install with Tessl CLI

npx tessl i github:NeverSight/skills_feed --skill skill-creator
What are skills?

81

Does it follow best practices?

Validation for skill structure

SKILL.md
Review
Evals

Evaluation results

100%

20%

Build a PDF Processing Skill

Skill initialization and packaging workflow

Criteria
Without context
With context

Used init_skill.py

100%

100%

Used package_skill.py

100%

100%

Packaged .skill file exists

100%

100%

Frontmatter fields only

100%

100%

No auxiliary docs

100%

100%

Script tested

75%

100%

Example files removed

0%

100%

Script in scripts/ directory

100%

100%

Description includes trigger context

40%

100%

Imperative form in body

87%

100%

Without context: $0.6211 · 2m 2s · 34 turns · 39 in / 7,792 out tokens

With context: $0.9908 · 2m 40s · 47 turns · 84 in / 9,815 out tokens

100%

28%

Organize Cloud Deployment Documentation into a Skill

Progressive disclosure and reference file organization

Criteria
Without context
With context

SKILL.md under 500 lines

100%

100%

Provider-specific reference files

100%

100%

References one level deep

0%

100%

No content duplication

100%

100%

SKILL.md references all files

100%

100%

Table of contents in reference files

0%

100%

SKILL.md stays high-level

100%

100%

design-notes explains structure

100%

100%

Frontmatter fields only

0%

100%

No auxiliary docs

100%

100%

Without context: $0.4641 · 1m 58s · 22 turns · 27 in / 7,059 out tokens

With context: $0.7454 · 2m 28s · 29 turns · 145 in / 9,135 out tokens

95%

39%

Design a SKILL.md for an Internal Data Query Assistant

SKILL.md description design and content quality

Criteria
Without context
With context

Description includes trigger context

16%

100%

No 'when to use' section in body

20%

100%

Schemas in reference files

60%

100%

Domain-specific reference files

100%

100%

Imperative form in SKILL.md body

100%

100%

No content duplication

30%

80%

SKILL.md references files with guidance

100%

100%

Degrees of freedom matched

90%

90%

Frontmatter fields only

0%

100%

review.md addresses criteria

50%

83%

Without context: $0.3203 · 1m 50s · 15 turns · 68 in / 6,330 out tokens

With context: $0.5793 · 2m 9s · 23 turns · 457 in / 7,394 out tokens

Evaluated
Agent
Claude Code

Table of Contents

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.