OrodruinLabs/nazgul:enhance

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.