Skip to content
Back to skills

Enhance

ASecurity

Research latest Claude Code features and propose Nazgul improvements. Use periodically to keep Nazgul aligned with new platform capabilities. Run with `/loop 2w /nazgul:enhance` for auto-recurring checks.

  • 7 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added May 26, 2026
researchbashperformance

Works with

  • claude code
  • mcp

Security analysis

A100/100

Scanned May 27, 2026

npx -y skills add OrodruinLabs/nazgul --skill enhance --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Enhance?

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

Security grade badge for Enhance
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/orodruinlabs-enhance/badge)](https://www.skillsdirectory.com/skills/orodruinlabs-enhance)

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: nazgul:enhance
description: Research latest Claude Code features and propose Nazgul improvements. Use periodically to keep Nazgul aligned with new platform capabilities. Run with `/loop 2w /nazgul:enhance` for auto-recurring checks.
context: fork
allowed-tools: Read, Bash, Glob, Grep, WebSearch, WebFetch
metadata:
  author: Jose Mejia
  version: 1.2.2
---

# Nazgul Enhance

## Examples
- `/nazgul:enhance` — Research latest Claude Code features and propose Nazgul improvements
- `/loop 2w /nazgul:enhance` — Auto-check every 2 weeks for new enhancement opportunities

## Current Nazgul Capabilities

### Plugin Version
- Version: !`jq -r '.version // "unknown"' .claude-plugin/plugin.json 2>/dev/null || echo "unknown"`

### Registered Hooks
- Hook types: !`jq -r '.hooks | keys | join(", ")' hooks/hooks.json 2>/dev/null || echo "unknown"`

### Skills Inventory
- Skills: !`ls -1 skills/ 2>/dev/null | tr '\n' ', ' | sed 's/,$//'`

### Agents Inventory
- Pipeline agents: !`ls -1 agents/*.md 2>/dev/null | xargs -I{} basename {} .md | tr '\n' ', ' | sed 's/,$//'`
- Reviewer templates: !`ls -1 agents/templates/*.md 2>/dev/null | xargs -I{} basename {} .md | tr '\n' ', ' | sed 's/,$//'`

### Config Schema
- Schema version: !`jq -r '.schema_version // "unknown"' templates/config.json 2>/dev/null || echo "unknown"`
- Config sections: !`jq -r 'keys | join(", ")' templates/config.json 2>/dev/null || echo "unknown"`

### Hook Scripts
- Scripts: !`ls -1 scripts/*.sh 2>/dev/null | xargs -I{} basename {} .sh | tr '\n' ', ' | sed 's/,$//'`

### Agent Tools Used
- Implementer tools: !`sed -n '/^tools:/,/^[^ -]/p' agents/implementer.md 2>/dev/null | grep '  - ' | sed 's/  - //' | tr '\n' ', ' | sed 's/,$//'`
- Team orchestrator tools: !`sed -n '/^tools:/,/^[^ -]/p' agents/team-orchestrator.md 2>/dev/null | grep '  - ' | sed 's/  - //' | tr '\n' ', ' | sed 's/,$//'`

## Instructions

Format all output per `references/ui-brand.md` — use stage banners, status symbols, and display patterns defined there.

### Phase 1: Inventory Current Capabilities

Using the preprocessor data above, build an internal capability map:
- Which hook lifecycle events does Nazgul handle? (Stop, PreCompact, PreToolUse, PostToolUse, SessionStart, SessionEnd — anything missing?)
- Which Claude Code tools do agents currently use? (Check allowed-tools in agent frontmatter)
- What config options exist? (Check templates/config.json sections)
- What scripts handle automation? (Check scripts/ directory)

### Phase 2: Research Claude Code Releases

Search for the latest Claude Code features and changes. Use these searches:

1. `WebSearch` for "Claude Code changelog 2026" and "Claude Code new features"
2. `WebSearch` for "Claude Code hooks reference" and "Claude Code agent teams"
3. `WebFetch` the official docs if URLs are found
4. `WebSearch` for "Anthropic Claude Code release notes"

Focus on:
- New hook types (e.g., PostCompact, Elicitation hooks, HTTP hooks)
- New tools available to agents (e.g., ExitWorktree, EnterWorktree)
- New configuration options (e.g., worktree sparse paths, session naming)
- New agent/skill capabilities (e.g., /loop, /plan with description)
- Performance features (e.g., fast mode, server-side compaction)
- New MCP integrations

### Phase 3: Gap Analysis

Compare discovered features against the capability inventory. For each gap, classify impact:

| Impact | Criteria |
|--------|----------|
| HIGH   | Improves core loop reliability OR enables capability Nazgul can't do today |
| MEDIUM | Improves performance/UX without new capability |
| LOW    | Already handled by Nazgul's own implementation OR not applicable |

### Phase 4: Output Enhancement Proposal

Output a structured report:

```
─── ◈ NAZGUL ▸ ENHANCING ─────────────────────────────

Enhancement Proposals
═══════════════════════════════════════

Current State
─────────────────────────────────────
Plugin version:    [version]
Schema version:    [version]
Hook types:        [count] registered
Skills:            [count] active
Agents:            [count] defined

New Features Discovered
─────────────────────────────────────

◆ [HIGH] Feature Name
  Source:    Claude Code vX.Y.Z (YYYY-MM-DD)
  Status:   Not implemented / Partially implemented
  Proposal: [What to do]
  Files:    [Which Nazgul files to modify]
  Effort:   Trivial / Moderate / Significant

◆ [MEDIUM] Feature Name
  ...

◇ [LOW] Feature Name
  ...

─── ◈ NEXT ─────────────────────────────────────────────
  Run `/nazgul:start "integrate [feature]"` to implement
  Or `/loop 2w /nazgul:enhance` for recurring checks
────────────────────────────────────────────────────────
```

### Important Notes

- This skill is READ-ONLY. It does NOT modify Nazgul files.
- It produces proposals for the user to act on.
- When run via `/loop`, it auto-checks periodically without user intervention.
- Focus on actionable, specific proposals — not vague suggestions.
- Always check if a feature is already handled by Nazgul before proposing it.

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.