Evaluates Claude Agent Skills on 10 quality axes with letter grades (A+ through F) and specific improvement recommendations. Use when auditing a skill, comparing skills, prioritizing improvements, or performing quality control on a skill library. Activate on "grade skill", "evaluate skill", "skill quality", "skill audit", "skill review", "rate skill". NOT for creating skills (use skill-architect), grading code quality, or evaluating non-skill documents.
Standardmäßig ist der Prompt ausgewählt, der zuerst die Quelle prüft. Sie können zu einem direkten Befehl wechseln oder eine lokale Kopie herunterladen.
Quelldateien prüfen
Lesen Sie SKILL.md und alle von SkillsMP angezeigten Begleitdateien, bevor Sie sich für eine Installation entscheiden.
Mit Codex oder Claude installieren Kopieren Sie diesen Prompt, fügen Sie ihn in Codex, Claude oder einen anderen Assistant ein und lassen Sie die Skill-Seite prüfen und installieren.
Ein direkter Befehl überspringt den Prüf-Prompt. Prüfen Sie die Quelle, bevor Sie ihn ausführen.
Evaluates Claude Agent Skills on 10 quality axes with letter grades (A+ through F) and specific improvement recommendations. Use when auditing a skill, comparing skills, prioritizing improvements, or performing quality control on a skill library. Activate on "grade skill", "evaluate skill", "skill quality", "skill audit", "skill review", "rate skill". NOT for creating skills (use skill-architect), grading code quality, or evaluating non-skill documents.
allowed-tools
Read,Grep,Glob
argument-hint
[skill-path]
metadata
{"category":"Productivity & Meta","tags":["grader","grade-skill","evaluate-skill","skill-quality"],"pairs-with":[{"skill":"skill-architect","reason":"Grading reveals architectural gaps that the architect addresses with structural improvements"},{"skill":"skill-creator","reason":"Grading feedback guides creators on which quality axes need improvement"},{"skill":"skill-coach","reason":"Grading provides objective scores that coaching uses to prioritize improvement areas"}]}
Structured evaluation rubric for Claude Agent Skills. Produces letter grades (A+ through F) on 10 axes plus an overall grade, with specific improvement recommendations for each axis.
Designed for sub-agents and non-expert reviewers who need a mechanical, repeatable process for assessing skill quality without deep domain expertise.
When to Use
✅ Use for:
Auditing a single skill's quality
Comparing skills against each other
Prioritizing which skills to improve first
Quality control sweeps across a skill library
Generating improvement roadmaps
❌ NOT for:
Creating new skills (use skill-architect)
Grading code quality or non-skill documents
Evaluating agent performance (different from skill quality)
Grading Process
flowchart TD
A[Read SKILL.md + all files] --> B[Score each of 10 axes]
B --> C[Assign letter grade per axis]
C --> D[Compute overall grade]
D --> E[Write improvement recommendations]
E --> F[Produce grading report]
Step-by-Step
Read the entire skill folder — SKILL.md, all references, scripts, CHANGELOG, README
Score each axis — Use the rubric below (0-100 per axis)
Convert to letter grade — See grade scale
Compute overall grade — Weighted average (Description and Scope are 2x weight)
Write 1-3 specific improvements per axis scoring below B+
Produce the grading report in the output format below
The 10 Evaluation Axes
Axis 1: Description Quality (Weight: 2x)
Does the description follow [What] [When] [Keywords]. NOT for [Exclusions]?
Grade
Criteria
A
Specific verb+noun, domain keywords users would type, 2-5 explicit NOT exclusions, 25-50 words
B
Has keywords and NOT clause, but slightly vague or missing synonym coverage
C
Too generic, missing NOT clause, or >100 words of process detail
D
Single vague sentence ("helps with X") or name/description mismatch
F
Missing or empty description
Axis 2: Scope Discipline (Weight: 2x)
Is the skill narrowly focused on one expertise type, or a catch-all?
Grade
Criteria
A
One clear expertise domain, "When to Use" and "NOT for" sections both present and specific
B
Mostly focused, minor boundary ambiguity
C
Covers 2-3 related but distinct domains, should probably be split
D
Catch-all skill ("helps with anything related to X")
F
No scope boundaries defined at all
Axis 3: Progressive Disclosure
Does the skill follow the three-layer architecture (metadata → SKILL.md → references)?
Grade
Criteria
A
SKILL.md <300 lines, heavy content in references, reference index in SKILL.md with 1-line descriptions
B
SKILL.md <500 lines, some references used, index present
C
SKILL.md >500 lines, or all content inlined with no references
D
SKILL.md >800 lines, or references exist but aren't indexed in SKILL.md
F
Single massive file with no structure
Axis 4: Anti-Pattern Coverage
Does the skill encode expert knowledge that prevents common mistakes?
Grade
Criteria
A
3+ anti-patterns with Novice/Expert/Timeline template, LLM-mistake notes
B
1-2 anti-patterns with clear explanation
C
Anti-patterns mentioned but no structured template
D
No anti-patterns, just positive instructions
F
Contains advice that IS an anti-pattern (outdated, harmful)
Axis 5: Self-Contained Tools
Does the skill include working tools (scripts, MCPs, subagents)?
Grade
Criteria
A
Working scripts with CLI interface, error handling, dependency docs; OR valid "no tools needed" justification
B
Scripts exist and work but lack error handling or docs
C
Scripts referenced but are templates/pseudocode
D
Phantom tools (SKILL.md references files that don't exist)
F
References non-existent tools AND no acknowledgment
Note: Not every skill needs tools. A pure decision-tree skill can score A if tools aren't applicable.
Axis 6: Activation Precision
Would the skill activate correctly on relevant queries and stay silent on irrelevant ones?
Grade
Criteria
A
Description has specific keywords matching user language, clear NOT clause, no obvious false-positive vectors
B
Good keywords, minor false-positive risk
C
Generic keywords that overlap with other skills
D
No specific keywords, or NOT clause contradicts intended use
F
Description would cause constant false activation
Axis 7: Visual Artifacts
Does the skill use Mermaid diagrams, code examples, and tables effectively?
Grade
Criteria
A
Decision trees as Mermaid flowcharts, tables for comparisons, code examples for concrete patterns
B
Some diagrams or tables, but key decision trees still in prose
C
Tables used but no Mermaid diagrams for processes
D
Prose-only, no visual structure
F
Wall of text with no formatting aids
Axis 8: Output Contracts
Does the skill define what it produces in a format consumable by other agents?
Grade
Criteria
A
Explicit output format (JSON schema, markdown template, or structured sections), subagent-consumable
B
Output format implied but not explicitly documented
C
No output format, but content is structured enough to infer
D
Unstructured prose output expected
F
N/A (pure reference skill) — exempt from this axis
Axis 9: Temporal Awareness
Does the skill track when knowledge was current and what has changed?
Grade
Criteria
A
Timelines in anti-patterns, "as of [date]" markers, CHANGELOG with dates
B
Some temporal context, CHANGELOG exists
C
No dates on knowledge, but CHANGELOG exists
D
No temporal context anywhere, knowledge could be stale
F
Contains demonstrably outdated advice without warning
Axis 10: Documentation Quality
README, CHANGELOG, and reference organization.
Grade
Criteria
A
README with quick start, CHANGELOG with dated versions, references well-organized with clear filenames
B
README and CHANGELOG exist, references present
C
SKILL.md is the only file, but it's well-structured
D
No README, no CHANGELOG, disorganized references
F
SKILL.md is the only file and it's poorly structured
Grade Scale
Letter
Score Range
Meaning
A+
97-100
Exemplary — sets the standard
A
93-96
Excellent — minor improvements possible
A-
90-92
Very good — a few small gaps
B+
87-89
Good — notable room for improvement
B
83-86
Solid — several areas need work
B-
80-82
Above average — meaningful gaps
C+
77-79
Average — significant improvements needed
C
73-76
Below average — major gaps
C-
70-72
Barely adequate
D+
67-69
Poor — fundamental issues
D
63-66
Very poor — needs major rework
D-
60-62
Near-failing quality
F
<60
Failing — start over
Overall Grade Computation
Axes 1 (Description) and 2 (Scope) carry 2x weight. All others carry 1x weight. If Axis 8 (Output Contracts) is marked exempt, remove it from the calculation.
Wrong: Giving B+ because "it's pretty good" without checking criteria.
Right: Match observations to the rubric table literally. If the description lacks a NOT clause, it cannot score above C on Axis 1.
Missing Context
Wrong: Grading a pure decision-tree skill poorly on Axis 5 (tools) because it has no scripts.
Right: Mark Axis 5 as "A — tools not applicable for this skill type."
Ignoring Phantoms
Wrong: Scoring Axis 5 as B because scripts are "referenced."
Right: Actually check if every referenced file exists. If scripts/validate.py is mentioned but doesn't exist, that's D.
Output Contract
This skill produces:
10-axis grade report with letter grades (A+ through F) and weighted total
Per-axis feedback with specific evidence and improvement suggestions
Priority improvements ranked by impact on overall quality score
Before/after comparison projecting grade changes from recommended fixes
