nelson
Orchestrates complex multi-agent software development using a structured Royal Navy squadron metaphor, featuring mission planning, parallel task coordination, and rigorous audit logs.
Introduction
Nelson is a comprehensive skill for Claude Code designed to manage complex, multi-agent development missions. It replaces chaotic agent execution with a disciplined, operational framework inspired by naval command structures. By defining clear sailing orders, conduct of estimates, and battle plans, it ensures that your agent teams operate with high reliability, auditability, and progress transparency. It is the ideal tool for developers tackling multi-system refactors, large-scale feature migrations, or complex codebase maintenance that requires sustained, multi-day coordination.
-
Multi-agent coordination: Orchestrates parallel workstreams via sub-agents or agent teams, ensuring that tasks are decomposed, assigned, and managed through standardized progress checkpoints.
-
Risk-gated execution: Implements four distinct risk tiers, from Patrol to Trafalgar, ensuring that high-risk code changes always require human oversight while low-risk tasks proceed with minimal friction.
-
Decision logging and auditability: Automatically generates captain’s logs, damage reports, and turnover briefs, providing a permanent, searchable record of every decision, intervention, and state change within a mission.
-
Damage control protocols: Features ten pre-defined, rigorous procedures to handle common agent pitfalls such as context window exhaustion, faulty output, budgetary overruns, or stalled execution.
-
Structured mission lifecycle: Enforces a formal progression from sailing orders and tactical estimates to battle plans and final stand-down, ensuring no mission component is overlooked.
-
Use this skill when your project involves multi-file refactoring, dependency migrations, or complex logic changes that demand a documented decision-making trail.
-
The workflow begins by establishing a dedicated mission directory (.nelson/missions/) that acts as the single source of truth for the session state and generated artifacts.
-
Inputs include the mission goal, constraints, and success metrics; outputs include structured JSON logs, progress reports, and the final state of your codebase.
-
It is highly recommended to conduct the Seven Question Maritime Tactical Estimate (reconnaissance, intent, effects, terrain, forces, coordination, and control) for any mission that spans multiple subsystems to ensure clear alignment before code generation starts.
-
Nelson is designed for complex, high-stakes development; for simple, single-file edits or isolated scripts, consider using more lightweight approaches.
Repository Stats
- Stars
- 335
- Forks
- 30
- Open Issues
- 13
- Language
- Python
- Default Branch
- main
- Sync Status
- Idle
- Last Synced
- May 3, 2026, 06:33 PM