AI Agentsagent-coordinationai-agentsanthropicclaude-codeclaude-code-skill
Installation
# Add to your Claude Code skills
git clone https://github.com/harrymunro/nelson
README.md
Nelson
Squadron-scale agent coordination for Claude Code — with risk tiers, damage control, and decision logs.
A Claude Code skill that organises multi-agent work into structured naval operations: sailing orders define the mission, captains command parallel workstreams, action stations enforce risk-appropriate controls, and a captain's log captures every decision for audit.
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Opencode, Cursor and beyond.
Use Nelson to migrate the payment module from Stripe v2 to v3
Nelson is a Claude Code skill — it loads automatically when your request matches. No slash command needed. See Prerequisites for the full agent-team experience with split panes.
Nelson gives Claude a six-step operational framework for tackling complex missions:
Sailing Orders — Define the outcome, success metric, constraints, and stop criteria
Form the Squadron — Choose an execution mode (single-session, subagents, or agent team) and size the team
Battle Plan — Split the mission into independent tasks with owners, dependencies, and file ownership
Quarterdeck Rhythm — Run checkpoints to track progress, identify blockers, monitor hull integrity, and manage budget
Action Stations — Classify tasks by risk tier and enforce verification before marking complete
Stand Down — Produce a captain's log with decisions, artifacts, validation evidence, and follow-ups
Why Nelson?
Most agent orchestration tools focus on starting missions. Nelson focuses on completing them safely.
Nelson gives your missions a shared vocabulary: "action stations" instead of "risk tier escalation", "hull integrity" instead of "context window consumption", "man overboard" instead of "stuck agent replacement". The names stick. So do the habits.
Risk-gated execution — Four station tiers (Patrol through Trafalgar) classify every task before it runs. High-risk work requires human confirmation; low-risk work flows without ceremony.
Damage control built in — Ten named procedures for stuck agents, context exhaustion, faulty output, budget overruns, and mission abort. These are protocols, not improvisation.
A decision log by default — Captain's log, quarterdeck reports, and turnover briefs are written as the mission runs. Every decision is auditable after the session ends.
Nelson coordinates its own development — the v1.7.0 release was planned and executed as a Nelson mission.
Who is this for?
You run Claude Code missions spanning multiple files or modules in parallel
You want structured checkpoints, risk classification, and a decision log
You've lost work to context exhaustion and want systematic handover procedures
You care about auditability — knowing what was decided, by which agent, and why
It may be overkill if you're doing a quick, single-file edit.
How Nelson compares
Both rapid-execution frameworks and Nelson's structured approach are useful — they optimise for different constraints.
| Approach | Best when | Trade-off |
|---|---|---|
| Nelson Navy structure | You need repeatable quality gates, explicit ownership, and a clear decision log across parallel work | More setup and coordination overhead up front |
| OmO/RuFlo-style rapid flow | You need the fastest possible movement on a narrow, low-risk path | Less formal checkpointing and role separation |
If you need fast parallel execution with minimal ceremony, OmO or RuFlo may suit you better. If coordination, auditability, and safe scaling matter more than raw tempo, Nelson is the better fit.
How it works
Execution modes
The skill selects one of three execution modes based on your mission:
| Mode | When to use | How it works |
|------|------------|--------------|
| single-session | Sequential tasks, low complexity, heavy same-file editing | Claude works through tasks in order within one session |
| subagents | Parallel tasks where workers only report back to the coordinator | Claude spawns subagents that work independently and return results |
| agent-team | Parallel tasks where workers need to coordinate with each other | Claude creates an agent team with direct teammate-to-teammate communication |
Chain of command
Nelson uses a three-tier hierarchy. The admiral coordinates captains, each captain commands a named ship, and crew members aboard each ship do the specialist work.
┌───────────┐
│ Admiral │
└─────┬─────┘
┌─────────────┼─────────────┐
▼ ▼ ▼
┌───────────┐ ┌───────────┐ ┌───────────┐
│ Captain │ │ Captain │ │ Red-Cell │
│ HMS Daring │ │ HMS Kent │ │ Navigator │
└─────┬─────┘ └─────┬─────┘ └───────────┘
┌────┼────┐ ┌────┼────┐
▼ ▼ ▼ ▼ ▼ ▼
XO PWO MEO PWO NO COX
Squadron level:
Admiral — Coordinates the mission, delegates tasks, resolves blockers. Coordinates final synthesis but does not perform it directly. There is always exactly one.
Captains — Each commands a named ship. Breaks their task into sub-tasks, crews specialist roles, coordinates crew, and verifies outputs. Implements directly only when the task is atomic. Typically 2-7 per mission.
Red-cell navigator — Challenges assumptions, validates outputs, and checks rollback readiness. Added for medium/high risk work.
Ship level (crew per captain, 0-4 members):
| Role | Abbr | Function | When to crew |
|------|------|----------|-------------|
| Executive Officer | XO | Integration & orchestration | 3+ crew or interdependent sub-tasks |
| Principal Warfare Officer | PWO | Core implementation | Almost always (default doer) |
| Navigating Officer | NO | Codebase research & exploration | Unfamiliar code, large codebase |
| Marine Engineering Officer | MEO | Testing & validation | Station 1+ or non-trivial verification |
| Weapon Engineering Officer | WEO | Config, infrastructure & systems integration | Significant config/infra work |
| Logistics Officer | LOGO | Documentation & dependency management | Docs as deliverable, dep management |
| Coxswain | COX | Standards review & quality | Station 1+ with established conventions |
Navigating Officer (NO) and Coxswain (COX) are read-only — they report findings but never modify files.
Ships are named from real Royal Navy warships, matched roughly to task weight: frigates for general-purpose, destroyers for high-tempo, patrol vessels for small tasks, historic flagships for critical-path, and submarines for research.
Squadron size caps at 10 squadron-level agents (admiral, captains, red-cell navigator). Crew are additional — up to 4 per ship. If a task needs more crew, split it into two ships.
Here's an example of the crew in action: we create four captains. Two of the captains are single-crew (minimum, since captains don't do the work themselves) and two of them are two-crew. So we have 11 agents working together in total:
