agentleFS
ROCmROCm/rocprofiler-systems-skills/skills/radisha-create-skill/SKILL.md

radisha-create-skill

Skill5 starsChanged 17 days agoMITView on GitHub

Use when creating a new skill for radisha - validates content, checks for conflicts, and integrates with existing skills

The file

Show its rules and text

We split this skill into the rules it gives an agent

14 rules. About 0 are common, 12 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈12

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this skill when:

    line 12
  2. Distinctive:

    When user provides skill content, analyze these aspects:

    line 67
  3. Distinctive:

    Check how the skill fits with existing radisha skills:

    line 82
  4. Distinctive:

    Check for conflicts with existing skills:

    line 93
  5. Distinctive:

    Verify the skill follows radisha patterns:

    line 113
  6. Distinctive:

    Wait for user approval before proceeding.

    line 162
  7. Distinctive:

    Ensure skill follows this structure:

    line 183
  8. Distinctive:

    Before creating, apply all fixes identified in Phase 2:

    line 233
  9. Distinctive:

    Add checkpoints if missing

    line 238
  10. Very specific:

    Write SKILL.md to the directory.

    line 247
  11. Distinctive:

    After creating the skill, update these files:

    line 251
  12. Distinctive:

    Add to appropriate section:

    line 255
  13. Distinctive:

    If conflicts were found, update related skills to disambiguate:

    line 290
  14. Very specific:

    Example: When creating exploration-explore-code, we updated ask skill:

    line 292

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Create Skill

Meta-skill for creating new radisha skills. Validates proposed skill content, checks for conflicts with existing skills, and handles integration.

When to Use

Use this skill when:

  • User provides content for a new skill
  • User asks to create a skill for a specific purpose
  • Converting existing documentation into a skill

Process Overview

┌─────────────────────────────────────────────────────────────────┐
│                  User provides skill content                     │
└─────────────────────────────────────────────────────────────────┘
                                │
                                ▼
                    ┌───────────────────────┐
                    │ Phase 1: Analyze      │
                    │ - Usefulness          │
                    │ - Alignment           │
                    │ - Conflicts           │
                    └───────────────────────┘
                                │
                                ▼
                    ┌───────────────────────┐
                    │ Phase 2: Report       │
                    │ Present findings to   │
                    │ user, suggest fixes   │
                    └───────────────────────┘
                                │
                                ▼
                    ┌───────────────────────┐
                    │ User approves?        │
                    └───────────────────────┘
                          │           │
                         no          yes
                          │           │
                          ▼           ▼
                    ┌─────────┐ ┌───────────────────────┐
                    │  Stop   │ │ Phase 3: Create skill │
                    └─────────┘ └───────────────────────┘
                                          │
                                          ▼
                              ┌───────────────────────┐
                              │ Phase 4: Integrate    │
                              │ - Update radisha-help │
                              │ - Update README       │
                              │ - Update related skills│
                              └───────────────────────┘
                                          │
                                          ▼
                              ┌───────────────────────┐
                              │ Report completion     │
                              └───────────────────────┘

Phase 1: Analyze Skill Content

When user provides skill content, analyze these aspects:

1.1 Usefulness Assessment

Answer: Will this skill help the AI perform better?

QuestionGood SignBad Sign
Does it provide structure?Clear phases, checkpointsVague guidelines
Does it reduce ambiguity?Explicit triggers, outputs"Use when appropriate"
Does it prevent mistakes?Common mistakes sectionNo anti-patterns
Is it actionable?Specific steps to followAbstract principles only

1.2 Alignment Check

Check how the skill fits with existing radisha skills:

RelationshipCheck
PrecursorDoes it come before other skills? (e.g., exploration → planning)
SuccessorDoes it come after other skills? (e.g., planning → programming)
ParallelDoes it work alongside other skills?
IndependentIs it standalone?

1.3 Conflict Detection

Check for conflicts with existing skills:

For each existing skill:
  - Does the new skill overlap in purpose?
  - Are there contradictory instructions?
  - Is there ambiguity about when to use which?

Common conflicts to check:

Existing SkillPotential Conflict
askQuick explanations vs. structured analysis
planning-*Which planning type applies?
programming-*Language-specific overlap?
testing-*Testing approach conflicts?

1.4 Consistency Check

Verify the skill follows radisha patterns:

PatternCheck
Tool namesUses AskUserQuestion not AskQuestion
Task trackingReferences platform's task tools
Plan modePlatform-agnostic language
Output locationFollows planning/ convention if applicable
FormattingTables, ASCII diagrams, not prose-heavy
CheckpointsUser validation at key points

Phase 2: Report Findings

Present analysis to user in this format:

## Skill Analysis: [Skill Name]

### Usefulness
| Aspect | Assessment |
|--------|------------|
| Structure | [Clear/Vague] |
| Triggers | [Explicit/Ambiguous] |
| Anti-patterns | [Present/Missing] |
| Actionability | [High/Medium/Low] |

**Verdict:** [Useful / Needs work / Not recommended]

### Alignment
- **Relationship:** [Precursor to X / Successor to Y / Independent]
- **Workflow position:** [Where it fits in the skill chain]

### Conflicts
| Existing Skill | Conflict | Resolution |
|----------------|----------|------------|
| [skill] | [overlap description] | [how to fix] |

### Issues to Fix
1. [Issue 1 with suggested fix]
2. [Issue 2 with suggested fix]

### Recommended Category
`skills/[category]/[skill-name]/SKILL.md`

---

**Proceed with creation?** (with fixes applied)

Wait for user approval before proceeding.

Phase 3: Create Skill File

3.1 Determine Location

Choose category based on skill purpose:

PurposeCategory prefixPath
Planning workflowsplanning-skills/planning-[name]/
Language-specific codingprogramming-skills/programming-[lang]/
Testing approachestesting-skills/testing-[name]/
Code explorationexploration-skills/exploration-[name]/
Library-specificlibraries-skills/libraries-[name]/
Git workflowsgit-skills/git-[name]/
Radisha managementradisha-skills/radisha-[name]/
Standalone utilityTop-levelskills/[name]/

3.2 Apply Template

Ensure skill follows this structure:

---
name: category/skill-name
description: One-line description - when to use this skill
---

# Skill Title

Brief overview (1-2 sentences).

<IMPORTANT>
Critical rules that must always be followed.
</IMPORTANT>

## When to Use

Clear triggers for invoking this skill:
- Trigger 1
- Trigger 2

## Process / Phases

Step-by-step workflow with:
- Numbered phases
- Checkpoints for user validation
- Clear deliverables

## Output

What gets produced:
- File location (if applicable)
- Format/structure

## Common Mistakes

| Mistake | Fix |
|---------|-----|
| [anti-pattern] | [correct approach] |

## Integration with Other Skills

| After This Skill | Use |
|------------------|-----|
| [scenario] | [next skill] |

3.3 Apply Fixes

Before creating, apply all fixes identified in Phase 2:

  • Fix tool names
  • Add missing sections
  • Convert diagrams to ASCII
  • Add checkpoints if missing
  • Standardize formatting

3.4 Create Directory and File

mkdir -p skills/[category]/[skill-name]

Write SKILL.md to the directory.

Phase 4: Integrate with Radisha

After creating the skill, update these files:

4.1 Update radisha-help/SKILL.md

Add to appropriate section:

**[Category] skills:**
- `category/skill-name` - Brief description

4.2 Update README.md

Three updates required:

A. Skill Categories table:

### [Category] Skills

| Skill | Description |
|-------|-------------|
| `category/skill-name` | Description |

B. Detailed section:

#### `category/skill-name`
[Detailed description with key features]

C. Directory structure:

├── [category]/
│   └── [skill-name]/
│       └── SKILL.md

4.3 Update Related Skills (if needed)

If conflicts were found, update related skills to disambiguate:

Example: When creating exploration-explore-code, we updated ask skill:

## When NOT to Use
- **Systematic code exploration** → use `exploration-explore-code`

Phase 5: Report Completion

After all integration:

## Skill Created: [category/skill-name]

**Files created:**
- `skills/[category]/[skill-name]/SKILL.md`

**Files updated:**
- `skills/radisha-help/SKILL.md` - Added to skill list
- `README.md` - Added to tables and directory structure
- [Other updated files if any]

**Integration:**
- Workflow position: [where it fits]
- Related skills: [how it connects]

The skill is now ready to use.

Validation Checklist

Before completing, verify:

  • Skill file created in correct location
  • Frontmatter has correct name and description
  • Tool names are platform-correct (AskUserQuestion, etc.)
  • No conflicts with existing skills (or conflicts resolved)
  • radisha-help updated with skill reference
  • README.md tables updated
  • README.md detailed section added
  • README.md directory structure updated
  • Related skills updated if disambiguation needed

Common Mistakes

MistakeFix
Creating without analysisAlways do Phase 1-2 first
Skipping conflict checkRead existing skills before creating
Wrong tool namesUse AskUserQuestion, TaskCreate, etc.
Forgetting README updatesUpdate all three sections
Forgetting radisha-help updateAdd skill reference to radisha-help/SKILL.md
No user approvalAlways present analysis, wait for approval
Wrong categoryMatch purpose to existing categories

Skill Categories Reference

Category prefixPurposeExamples
planning-Task planning workflowsplanning-feature, planning-bugfix, planning-refactor
programming-Language-specific codingprogramming-cpp, programming-python, programming-cmake-best-practices
testing-Testing approachestesting-testplan, testing-gtest-gmock, testing-pytest
exploration-Code understandingexploration-explore-code
libraries-Library-specific knowledgelibraries-amd-smi
git-Git workflowsgit-prepare-pull-request, pr-review
radisha-Radisha managementradisha-update, radisha-create-skill
askQuick explanations(standalone)

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.