dojo.md
University for AI agents. 92 courses, 4400+ scenarios, any model via OpenRouter. Auto-training loops generate per-model SKILL.md documents. Works with Claude Code, OpenClaw, Cursor, Windsurf. No fine-tuning req...
Use when optimizing existing Claude skills, checking skill quality, auditing skill compliance, or when skills have obvious issues. Triggers on skill optimization requests, quality checks, or skill improvement tasks.
Follow the repository instructions for this skill and your agent. Review the files and dependencies before adding it to your workflow.
SkillHub CLI
skillhub install acautomata/skill-improvementInspect the source and compatibility requirements, then follow its installation steps.
Give your agent the source link and ask it to inspect instructions and dependencies before explaining the installation steps.
Systematically optimize skill quality through a diagnostic-report-select-execute-verify workflow. Ensure skills comply with Claude's official best practices for maximum effectiveness.
Core principle: If you didn't diagnose a skill, you don't know what to fix.
digraph workflow {
"1. Diagnose" [shape=box];
"2. Report" [shape=box];
"3. Select" [shape=box];
"4. Execute" [shape=box];
"5. Verify" [shape=box];
"1. Diagnose" -> "2. Report";
"2. Report" -> "3. Select";
"3. Select" -> "4. Execute";
"4. Execute" -> "5. Verify";
"5. Verify" -> "3. Select" [label="fail"];
}
Scan skill for quality issues across 4 categories.
Categories:
Process:
Output: Raw issue list
Detailed checklist: See diagnostic-checklist.md
Present findings in structured format.
Report structure:
# Skill Diagnostic Report: [name]
**Grade:** [A/B/C/D]
**Issues:** X total (Y high, Z medium, W low)
## High Priority (Y)
[Issues that prevent discovery or execution]
## Medium Priority (Z)
[Issues that impact quality or usability]
## Low Priority (W)
[Minor improvements]
For each issue include:
Report templates: See report-templates.md
User chooses which iss...
These signals come from indexed source analysis and may lag behind changes. Check the current source files.
# Skills Improvement
## Overview
Systematically optimize skill quality through a diagnostic-report-select-execute-verify workflow. Ensure skills comply with Claude's official best practices for maximum effectiveness.
**Core principle:** If you didn't diagnose a skill, you don't know what to fix.
---
## Workflow
```dot
digraph workflow {
"1. Diagnose" [shape=box];
"2. Report" [shape=box];
"3. Select" [shape=box];
"4. Execute" [shape=box];
"5. Verify" [shape=box];
"1. Diagnose" -> "2. Report";
"2. Report" -> "3. Select";
"3. Select" -> "4. Execute";
"4. Execute" -> "5. Verify";
"5. Verify" -> "3. Select" [label="fail"];
}
```
---
## Phase 1: Diagnose
Scan skill for quality issues across 4 categories.
**Categories:**
- **Metadata** (HIGH): name, description, keywords
- **Architecture** (MEDIUM): file structure, progressive disclosure
- **Text** (MEDIUM): conciseness, clarity, token efficiency
- **Code** (HIGH): error handling, dependencies, validation
**Process:**
1. Read SKILL.md and all referenced files
2. Apply diagnostic checklist (see references/diagnostic-checklist.md)
3. Record each issue with category, location, severity
**Output:** Raw issue list
**Detailed checklist:** See [diagnostic-checklist.md](references/diagnostic-checklist.md)
---
## Phase 2: Report
Present findings in structured format.
**Report structure:**
```markdown
# Skill Diagnostic Report: [name]
**Grade:** [A/B/C/D]
**Issues:** X total (Y high, Z medium, W low)
## High Priority (Y)
[Issues that prevent discovery or execution]
## Medium Priority (Z)
[Issues that impact quality or usability]
## Low Priority (W)
[Minor improvements]
```
**For each issue include:**
- Category and check ID
- Current state vs expected state
- Impact explanation
- Specific fix recommendation
- Reference to quality standard
**Report templates:** See [report-templates.md](references/report-templates.md)
---
## Phase 3: Select
User chooses which iss...University for AI agents. 92 courses, 4400+ scenarios, any model via OpenRouter. Auto-training loops generate per-model SKILL.md documents. Works with Claude Code, OpenClaw, Cursor, Windsurf. No fine-tuning req...
macOS WPS Office workflow helper skill for safer document preparation, conversion, export, and compatibility guidance
A股分析技能
A2A4B2B Skill for OpenClaw