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 Gemini CLI's capabilities with specialized knowledge, workflows, or tool integrations.
80
52%
Does it follow best practices?
Impact
100%
1.29xAverage score across 6 eval scenarios
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./packages/core/src/skills/builtin/skill-creator/SKILL.mdQuality
Discovery
57%Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.
The description adequately communicates when to use the skill with an explicit trigger clause, but falls short on specificity—it reads more like a tagline than a detailed capability list. The reference to 'Gemini CLI' is either an error or a cross-platform concern that undermines clarity, and the lack of concrete actions (e.g., generating frontmatter, structuring markdown, validating fields) makes it hard to distinguish from a generic documentation helper.
Suggestions
Add specific concrete actions such as 'generates YAML frontmatter, writes markdown skill instructions, validates skill structure, and creates description fields' to improve specificity.
Include more natural trigger terms users would say, such as 'SKILL.md', 'skill file', 'write a skill', 'skill template', or 'new skill definition'.
Verify the product name—'Gemini CLI' appears to be incorrect if this skill is for Claude; correct it to avoid confusion and improve distinctiveness.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | The description uses vague language like 'Guide for creating effective skills' and 'extends capabilities with specialized knowledge, workflows, or tool integrations.' It does not list concrete actions such as 'generates YAML frontmatter,' 'writes markdown instructions,' or 'validates skill structure.' | 1 / 3 |
Completeness | It does answer both 'what' (guide for creating effective skills that extend capabilities) and 'when' (when users want to create a new skill or update an existing skill), with an explicit 'Use when' equivalent clause. | 3 / 3 |
Trigger Term Quality | It includes some relevant keywords like 'create a new skill,' 'update an existing skill,' and 'skill' which users might naturally say. However, it references 'Gemini CLI' which seems like a mismatch (possibly should be Claude), and it misses variations like 'SKILL.md,' 'skill file,' 'write a skill,' or 'skill template.' | 2 / 3 |
Distinctiveness Conflict Risk | The skill is somewhat specific to skill creation, which is a distinct niche. However, the vague phrasing about 'specialized knowledge, workflows, or tool integrations' could overlap with general documentation or configuration skills, and the mention of 'Gemini CLI' rather than Claude creates confusion. | 2 / 3 |
Total | 8 / 12 Passed |
Implementation
47%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a comprehensive meta-skill for creating skills that provides a well-structured 7-step workflow with good examples and patterns. However, it significantly violates its own core principle of conciseness—it's verbose, explains concepts the AI already knows, and includes inline content that should be in reference files per its own progressive disclosure guidance. The actionability is moderate, with some concrete commands but too much descriptive content relative to executable guidance.
Suggestions
Cut the 'About Skills' and 'What Skills Provide' sections significantly—Gemini CLI doesn't need to be told what skills are when it's already using one. Move straight to the creation process.
Move the detailed descriptions of bundled resources (scripts/, references/, assets/) and progressive disclosure patterns into a separate reference file, keeping only a brief summary in SKILL.md.
Replace the descriptive 'Body' subsection under Step 4 with concrete, actionable guidance—e.g., a checklist of what the body must contain, with a brief template or example.
Trim redundant explanations throughout—e.g., the 'What to Not Include' section lists obvious items (README.md, CHANGELOG.md) that the AI can infer from the principle of minimalism.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill is extremely verbose at ~400+ lines, explaining many concepts that Gemini CLI already knows (what PDFs are, what scripts do, what assets are). It includes extensive explanations of basic concepts like 'what skills provide,' lengthy descriptions of file types, and redundant elaboration throughout. Much of this could be cut by 50%+ without losing actionable content. | 1 / 3 |
Actionability | The skill provides some concrete guidance like specific bash commands for init/package scripts and directory structure examples, but much of the content is descriptive rather than instructive. The actual SKILL.md writing guidance is relatively vague ('Write instructions for using the skill'), and many sections describe concepts rather than giving executable steps. | 2 / 3 |
Workflow Clarity | The 7-step skill creation process is clearly sequenced with explicit ordering ('Follow these steps in order, skipping only if there is a clear reason'). Each step has clear entry/exit criteria, and Step 5 includes validation before packaging. The workflow includes feedback loops in Step 7 for iteration. | 3 / 3 |
Progressive Disclosure | The skill references external files (references/workflows.md, references/output-patterns.md) and demonstrates progressive disclosure patterns well in its examples. However, the SKILL.md itself is monolithic—much of the detailed content about bundled resources, progressive disclosure patterns, and examples could be moved to reference files, following its own advice about keeping SKILL.md under 500 lines. | 2 / 3 |
Total | 8 / 12 Passed |
Validation
100%Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.
Validation — 11 / 11 Passed
Validation for skill structure
No warnings or errors.
- Repository
- google-gemini/gemini-cli
- Commit
9f76f34
- Reviewed
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.