Skip to content
Back to skills

Skill Curator

ASecurity

Use when adding, importing, reorganizing, renaming, or refactoring skills according to the Lean Unified Skills Architecture.

  • 24 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 3, 2026
ai-agentspythongobashnodetestingrefactoringfrontend

Works with

  • cli

Security analysis

A100/100

Scanned September 3, 2026

npx -y skills add J-StaR-Films-Studios/VibeCode-Protocol-Suite --skill skill-curator --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Skill Curator?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Skill Curator
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/j-star-films-studios-skill-curator/badge)](https://www.skillsdirectory.com/skills/j-star-films-studios-skill-curator)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: skill-curator
description: Use when adding, importing, reorganizing, renaming, or refactoring skills according to the Lean Unified Skills Architecture.
author: J StaR Films / Takomi
coauthored: J StaR Films / Takomi
version: 1.0.0
---

# Skill Curator

Automated workflow for ingesting, categorizing, standardizing, and wiring new or existing skills into the Takomi Skills Ecosystem without prompt bloat.

---

## 1. The 6-Step Ingestion Workflow

When the user gives you a new skill (or asks to organize, rename, or import one), execute these 6 steps in order:

```mermaid
flowchart TD
    A["1. Inspect & Deduplicate"] --> B["2. Placement Decision"]
    B --> C["3. Standardize Frontmatter"]
    C --> D["4. Update Parent Router"]
    D --> E["5. Cleanse & Unslop"]
    E --> F["6. Register & Verify"]
```

---

### Step 1: Inspect & Deduplicate
1. Read the incoming skill's instructions, code, and helper scripts.
2. Search existing suites (`assets/.agent/skills/`) to check for duplicate or overlapping capabilities:
   ```bash
   node scripts/verify_skills_integrity.cjs
   ```
3. If an existing skill already handles this domain, present a recommendation to the user to either **merge**, **upgrade**, or **co-locate** the skill.

---

### Step 2: Placement Decision
Apply the **Lean Unified Skills Architecture** rules:

| Condition | Destination | Action |
| :--- | :--- | :--- |
| Relates to an existing domain (frontend, video, dev, testing, agents, docs, marketing) | **Existing Umbrella Suite** | Place as a sub-folder inside the suite (e.g. `assets/.agent/skills/<suite>/<sub-skill>/`). |
| Single-purpose CLI binary wrapper or global mode | **Standalone Skill** | Place at top level `assets/.agent/skills/<skill>/`. |
| Distinct new domain with 3+ related sub-skills | **New Umbrella Suite** | Create new suite folder with top-level router `SKILL.md`. |
| **Core Essentials** | **Reserved** | **NEVER** automatically add to Core Essentials. Core status is strictly decided by the user. |

---

### Step 3: Standardize Frontmatter & Naming
Every `SKILL.md` MUST begin with standardized YAML frontmatter:

```yaml
---
name: my-skill-name
description: Use when [user wants to do X, trigger condition Y, or intent Z]... (Max 2 sentences, dense keyword triggers).
author: Original Author
coauthored: J StaR Films / Takomi
version: 1.0.0
---
```

**Naming Standards:**
- Directory name and `name` must be lowercase `kebab-case`.
- `description` must always begin with `"Use when..."` and contain specific, highly searchable domain keywords.

---

### Step 4: Update Parent Suite Router (If Sub-Skill)
When placing a sub-skill inside an umbrella suite:
1. **Update Router Frontmatter:** Add the sub-skill's keyword triggers to the parent suite's `SKILL.md` `description` field so harnesses index it in the prompt.
2. **Update Router Table:** Add a row to the sub-skills markdown table in the parent's `SKILL.md`:
   ```markdown
   | **`sub-skill-name`** | Brief one-line purpose | [`sub-skill-name/SKILL.md`](sub-skill-name/SKILL.md) |
   ```

---

### Step 5: Cleanse, Unslop & Platform Hardening
1. **Apply Unslop Filter:** Read `.agents/skills/unslop/SKILL.md`. Remove conversational AI filler, unparsed Markdoc/HTML tags (`{% callout %}`, `{% image %}`, `{% tabs %}`), and UI-specific button assumptions.
2. **Convert to Direct Instructions:** State what the agent should do directly and concisely.
3. **Windows UTF-8 Hardening (Python Scripts):** If the skill includes Python CLI scripts, ensure standard output handles UTF-8:
   ```python
   import sys
   if hasattr(sys.stdout, 'reconfigure'):
       sys.stdout.reconfigure(encoding='utf-8')
   if hasattr(sys.stderr, 'reconfigure'):
       sys.stderr.reconfigure(encoding='utf-8')
   ```

---

### Step 6: Register & Verify
1. **Catalog Registration:** If adding a top-level standalone skill or new suite, register it in `.pi/extensions/takomi-context-manager/skill-categories.js` under the appropriate category in `SKILL_CATEGORIES`.
2. **Run Integrity Verification:**
   ```bash
   node scripts/verify_skills_integrity.cjs
   ```
   Confirm that all relative links resolve and 0 errors are reported.
3. **Run Test Suite:**
   ```bash
   npm test
   ```

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.