[02] META. Create new skills when existing ones don't cover the task. Analyze unique requirements, build framework (Frame → Research → Plan → Execute), integrate risks, and declare new skill. Use when facing novel problems that existing skills can't address.
68
51%
Does it follow best practices?
Impact
96%
1.62xAverage score across 3 eval scenarios
Passed
No known issues
Optimize this skill with Tessl
npx tessl skill review --optimize ./skills/data/02-meta-skill-forge-150/SKILL.mdQuality
Discovery
75%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 its meta-purpose of creating new skills and includes an explicit 'Use when' clause, giving it good completeness and distinctiveness. However, the specificity of actions could be improved beyond the abstract framework acronym, and trigger terms could better match natural user language patterns for when they'd want to create a new skill.
Suggestions
Add more natural trigger terms users might say, such as 'teach you a new task', 'create a workflow', 'add a new capability', or 'no existing skill fits'.
Replace or supplement the abstract framework notation (Frame → Research → Plan → Execute) with more concrete output descriptions, e.g., 'generates a new SKILL.md file with structured instructions, triggers, and edge cases'.
| Dimension | Reasoning | Score |
|---|---|---|
Specificity | Names the domain (creating new skills) and some actions ('analyze unique requirements', 'build framework', 'declare new skill'), but the framework steps (Frame → Research → Plan → Execute) are somewhat abstract rather than concrete actions. It's more procedural than specific about what outputs are produced. | 2 / 3 |
Completeness | Clearly answers both what ('Create new skills when existing ones don't cover the task, analyze unique requirements, build framework, integrate risks, declare new skill') and when ('Use when facing novel problems that existing skills can't address') with an explicit trigger clause. | 3 / 3 |
Trigger Term Quality | Includes some relevant terms like 'novel problems', 'new skills', and 'existing skills can't address', but misses natural user phrases like 'teach you how to', 'learn a new task', 'create a workflow', or 'add a capability'. The term 'META' is jargon that users wouldn't naturally say. | 2 / 3 |
Distinctiveness Conflict Risk | This is a clearly distinct meta-skill about creating other skills, which is a unique niche unlikely to conflict with domain-specific skills. The '[02] META' prefix and the focus on 'novel problems that existing skills can't address' make it clearly distinguishable. | 3 / 3 |
Total | 10 / 12 Passed |
Implementation
27%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill is significantly over-engineered and verbose, ironically embodying the very anti-pattern it warns against ('Over-Engineering: Too complex for task'). The content is mostly abstract placeholder templates rather than concrete, actionable guidance, and the same information is repeated multiple times (e.g., the skill template appears twice). While the workflow structure is reasonable, the extreme length and redundancy make this skill a poor use of context window tokens.
Suggestions
Cut content by at least 60%: remove the duplicate skill template, collapse the 7 steps into concise descriptions, and eliminate sections that explain concepts Claude already understands (e.g., 'What This Skill Does' explaining what AI does).
Replace abstract placeholder templates with one concrete, complete worked example of forging a real skill, which would be far more actionable than fill-in-the-blank brackets.
Extract the full skill file template into a separate TEMPLATE.md file and reference it, rather than including it inline twice.
Remove the '150% Rule' framing throughout—it adds no actionable value and consumes significant tokens explaining a metaphor Claude doesn't need.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Extremely verbose at ~300+ lines. Massive amounts of template scaffolding and placeholder text that Claude doesn't need spelled out. The '150% Rule' table, 'What This Skill Does' section explaining what AI does, and extensive template boilerplate are all unnecessary padding. The same guidance could be conveyed in under 80 lines. | 1 / 3 |
Actionability | Provides structured templates with placeholder brackets (e.g., '[Unique aspect 1]', '[Challenge 1]') which give shape but are not executable or concrete. The examples section at the end provides some concrete guidance, but most of the skill is abstract fill-in-the-blank templates rather than specific, actionable instructions. | 2 / 3 |
Workflow Clarity | The 7-step sequence (Gap Analysis → Requirements → Framework Construction → Risk Integration → Quality Assurance → Skill Declaration → User Approval) is clearly ordered and includes validation checkpoints in Step 5. However, there's no feedback loop for when validation fails at intermediate steps, and the approval gate at Step 7 is the only real checkpoint. The workflow is more of a documentation template than an operational process with error recovery. | 2 / 3 |
Progressive Disclosure | Monolithic wall of text with everything inline. The full skill file template, all 7 execution steps with their complete templates, examples, operational rules, and failure modes are all crammed into one document. The template content in Step 6 and the 'Skill File Template' section are largely redundant. No references to external files for detailed content. | 1 / 3 |
Total | 6 / 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.
d156cd1
Table of Contents
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.