ClawKit Logo
ClawKitReliability Toolkit
Back to Registry
Official Verified

Citadel Agent Orchestration

Skill by adisinghstudent

skill-install — Terminal

Install via CLI (Recommended)

clawhub install openclaw/skills/skills/adisinghstudent/citadel-agent-orchestration
Or
---
name: citadel-agent-orchestration
description: Agent orchestration harness for Claude Code with four-tier routing, campaign persistence, parallel fleet agents, and 25 skills for autonomous coding campaigns.
triggers:
  - set up citadel for my project
  - run parallel agents with citadel
  - orchestrate a coding campaign
  - use citadel to build a feature
  - how do I use the /do router
  - run a fleet of agents in parallel
  - persist a coding campaign across sessions
  - automate code review with citadel
---

# Citadel — Agent Orchestration Harness for Claude Code

> Skill by [ara.so](https://ara.so) — Daily 2026 Skills collection.

Citadel is a Claude Code **plugin** that routes any coding task through the cheapest capable tool — from a direct one-line fix to a multi-day parallel agent campaign. It provides 25 skills, 4 autonomous sub-agents, 10 lifecycle hooks, and campaign persistence across sessions.

---

## Installation

**Prerequisites:** [Claude Code](https://docs.anthropic.com/en/docs/claude-code) + Node.js 18+

```bash
# 1. Clone Citadel once (global install — works across all projects)
git clone https://github.com/SethGammon/Citadel.git ~/citadel

# 2. Launch Claude Code with the plugin loaded
claude --plugin-dir ~/citadel

# 3. Run first-time setup inside any project
/do setup

Persistent install (recommended):

/plugin marketplace add ~/citadel
/plugin install citadel@citadel-local
/reload-plugins

After install, /do is available in every Claude Code session.


Core Concept: The /do Router

Type what you want in plain English. /do classifies intent and dispatches to the cheapest capable path automatically.

/do fix the typo on line 42          → Direct edit (zero overhead)
/do review the auth module           → /review (5-pass code review)
/do why is the API returning 500     → /systematic-debugging
/do add payments to my app           → /create-app tier 5
/do build me a recipe app            → /prd → /architect → /archon
/do overhaul all three services      → /fleet (parallel agents)
/do continue                         → resumes active campaign

You never have to choose the tier. If routing is wrong, say "wrong tool" or "just do it yourself" and it adjusts.


Four-Tier Orchestration Ladder

TierToolWhen to Use
1SkillSingle-domain task (review, test, doc, refactor)
2MarshalMulti-skill chain within one session
3ArchonMulti-session autonomous campaign with quality gates
4FleetParallel agents across isolated git worktrees

Key Skills Reference

App Creation

/prd "a recipe sharing app with user accounts and ratings"
# → Generates a full Product Requirements Document

/architect
# → Converts active PRD into file tree, build phases, end conditions

/create-app
# → End-to-end creation (5 tiers: blank → guided → templated → generated → feature addition)

Metadata

Stars3809
Views0
Updated2026-04-05
View Author Profile
AI Skill Finder

Not sure this is the right skill?

Describe what you want to build — we'll match you to the best skill from 16,000+ options.

Find the right skill
Add to Configuration

Paste this into your clawhub.json to enable this plugin.

{
  "plugins": {
    "official-adisinghstudent-citadel-agent-orchestration": {
      "enabled": true,
      "auto_update": true
    }
  }
}
Safety NoteClawKit audits metadata but not runtime behavior. Use with caution.