By LupusDei
Coordinate multiple Claude Code agents for epic planning, task execution, and inter-agent communication via tmux and MCP, with automated iOS TestFlight deployment using Fastlane and GitHub Actions.
Broadcast a message into every active agent's Claude Code tmux session as a submitted prompt. Also sends persistent MCP messages as backup. Use when you need to inject a standing order, instruction, or announcement into all running agents at once.
Send a direct message to a specific agent via MCP messaging and tmux injection. Guarantees delivery by injecting the message as a submitted prompt into the agent's tmux session AND sending a persistent MCP message. Use when you need to message a specific agent and get a response, rather than broadcasting to everyone.
Discuss and review an improvement proposal with the user. Analyzes strengths, weaknesses, feasibility, and cost estimate. Records the review as a persistent comment on the proposal, and optionally creates a revision if concrete improvements are identified. Use when the user says "discuss proposal", "review proposal", "what do you think about this proposal", or references a proposal ID for discussion and refinement rather than execution.
Create structured epic hierarchies for new features, major refactorings, and large tasks. Generates speckit-style artifacts (spec.md, plan.md, tasks.md, beads-import.md) in specs/ and creates real beads via `bd` CLI with wired dependencies. Use when the user says things like "Create a new epic for ...", "I want to do a major refactoring ...", "Time for a new plan", "Let's add a new feature", "New epic", or any request to plan and track a significant body of work.
Execute an accepted proposal by fetching it from the backend and orchestrating epic creation via the epic-planner skill. Use when the user says "execute proposal", "implement proposal", "start working on proposal X", or wants to turn a proposal into an actionable epic hierarchy with beads.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
/ˈajət(ə)nt/ — noun, Military. An officer who assists a commanding officer with administrative duties, relays orders to the field, tracks unit readiness, and ensures every element of the force is accounted for and moving in concert.
In the 18th century, no army moved without its adjutant. While generals drew battle plans and captains led charges, the adjutant was the nerve center — the officer who knew where every regiment was, who was wounded, who was idle, and what orders hadn't reached the front yet. They kept the war diary. They dispatched riders. They turned chaos into coordination.
Adjutant does the same thing for your AI agents.
It is a real-time command post for multi-agent orchestration: a retro terminal-themed dashboard where you can watch your agents work, read their reports, assign them missions, and track every task from inception to completion. Agents connect via MCP, report status like field officers checking in at HQ, and communicate through persistent messaging channels — all backed by Beads issue tracking and a SQLite message store.
┌─────────────────────────────────────────────────────────┐
│ ADJUTANT v2.x ◉ 4 agents online 13:42:07 │
├──────────┬──────────────────────────────────────────────┤
│ AGENTS │ kerrigan ● working "API contract tests" │
│ │ amon ● working "C4-002 storyboard" │
│ │ mengsk ◐ idle "awaiting orders" │
│ │ raynor ○ offline │
├──────────┼──────────────────────────────────────────────┤
│ BEADS │ adj-074 ██████░░░░ 60% API contracts │
│ │ C4-002 █████████░ 90% Storyboard pipe │
│ │ C4-003 ██░░░░░░░░ 20% Prompt enhance │
├──────────┼──────────────────────────────────────────────┤
│ COMMS │ [kerrigan] Phase 1 complete. 25 tests pass │
│ │ [amon] QA found 9 bugs, bugfix dispatched │
│ │ [you] Roger. Prioritize the P1 fixes. │
└──────────┴──────────────────────────────────────────────┘
Your agents are the army. You are the general. Adjutant keeps the war room running.
npm install -g adjutant
adjutant
The dashboard opens at http://localhost:4200.
adjutant # Start dashboard (frontend + backend + ngrok)
adjutant --no-tunnel # Disable ngrok tunnel
adjutant --port 8080 # Custom frontend port (default: 4200)
adjutant --api-port 8081 # Custom backend port (default: 4201)
adjutant init # Bootstrap Adjutant in a project
adjutant doctor # Check system health and prerequisites
adjutant prime # Output PRIME.md agent protocol to stdout
adjutant unhook # Remove Adjutant plugin hooks from Claude Code
adjutant --help # Show all options
One-time run without installing:
npx adjutant
bd in PATHbrew install ngrok
ngrok config add-authtoken <your-token>
Adjutant integrates with Claude Code via its plugin system, providing MCP tools, epic planning, messaging skills, and automatic PRIME.md injection on session start.
After installing adjutant globally, run adjutant init in any project directory:
adjutant init
This performs the following:
| Step | What it does |
|---|---|
~/.adjutant/PRIME.md | Creates a global default agent protocol file |
.adjutant/PRIME.md | Creates a local override in the current project |
.mcp.json | Creates/merges MCP server config for adjutant |
| Plugin marketplace | Registers LupusDei/adjutant via claude plugin marketplace add |
| Plugin install | Installs adjutant-agent plugin with user scope |
| Plugin enable | Enables the plugin in Claude Code settings |
| Legacy cleanup | Removes old manual hooks from ~/.claude/settings.json |
All steps are idempotent -- safe to run multiple times.
adjutant doctor
Checks file existence, plugin status, network health, and tool availability.
Once installed, the plugin automatically:
npx claudepluginhub lupusdei/adjutant --plugin adjutant-agentMemory compression system for Claude Code - persist context across sessions
Ultra-compressed communication mode. Cuts ~75% of tokens while keeping full technical accuracy by speaking like a caveman.
Standalone image generation plugin using Nano Banana MCP server. Generates and edits images, icons, diagrams, patterns, and visual assets via Gemini image models. No Gemini CLI dependency required.
Multi-model consensus engine integrating OpenAI Codex CLI, Gemini CLI, and Claude CLI for collaborative code review and problem-solving.
Streamline people operations — recruiting, onboarding, performance reviews, compensation analysis, and policy guidance. Maintain compliance and keep your team running smoothly.
Write feature specs, plan roadmaps, and synthesize user research faster. Keep stakeholders updated and stay ahead of the competitive landscape.