fleet

Coordinate parallel /ship runs by analyzing cross-feature conflicts and batching non-conflicting work into isolated worktrees.

10|2|Updated Mar 2, 2026
One-click install
npx skills add https://github.com/dixus/claudeframework --skill fleet-dixus
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: fleet
Source: https://github.com/dixus/claudeframework/tree/main/.claude/skills/fleet
Command: npx skills add https://github.com/dixus/claudeframework --skill fleet-dixus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Orchestrates the coordination of parallel PRD ship runs by analyzing cross-feature file conflicts and batching non-conflicting work into parallel worktrees, then merging results back into main with guardrails.

Core Features & Use Cases

  • Conflict-aware batching of PRD features for parallel /ship runs
  • Isolated worktrees per batch to prevent cross-feature interference
  • Automated merging back into main with conflict detection and graceful fallback
  • Supports --dry-run, --max-parallel, and --batch-only options for planning and execution

Quick Start

Start by decomposing your PRDs and running the fleet to batch and ship in parallel.

Frequently Asked Questions about fleet

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run parallel ship tasks for decomposed PRDs without file conflicts?

Parallel ship tasks for decomposed PRDs are coordinated by analyzing cross-feature file conflicts and batching non-conflicting work into isolated worktrees. The system builds a conflict graph, assigns batches via greedy coloring, and merges results back into main.

What is conflict-aware batching for parallel feature development?

Conflict-aware batching is a technique that analyzes file overlaps between decomposed PRD features, groups non-conflicting work into parallel execution batches, and uses isolated worktrees to prevent cross-feature interference during automated merging.

How do I orchestrate multi-feature PRD ships in isolated worktrees?

Multi-feature PRD ships are orchestrated by reading a decomposition manifest, building a conflict graph, assigning execution batches via greedy coloring, and coordinating per-feature agents in isolated worktrees with automated merging and status updates.

Can I limit the number of parallel worktrees during batched ship runs?

Yes, parallel worktrees can be limited using the --max-parallel option. You can also use --dry-run for planning and conflict analysis without execution, or --batch-only to stop after batch assignment.

Does automated merging back into main handle conflicts gracefully?

Automated merging back into main includes conflict detection with graceful fallback. If conflicts are detected during the merge process, the system falls back safely rather than forcing a problematic merge.

When should I use batch-only mode for parallel PRD orchestration?

Batch-only mode should be used when you want to review the batch assignment and conflict graph before execution. It stops the fleet workflow after assigning batches via greedy coloring, prior to launching per-feature agents in worktrees.