task-init-fleet

Decomposes a complex brief into parallel sub-tasks initialized by concurrent worker agents.

6|Updated Jul 10, 2026
One-click install
npx skills add https://github.com/Mozurok/fhorja.dev --skill task-init-fleet-mozurok
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: task-init-fleet
Source: https://github.com/Mozurok/fhorja.dev/tree/main/.claude/skills/task-init-fleet
Command: npx skills add https://github.com/Mozurok/fhorja.dev --skill task-init-fleet-mozurok

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Initializing a large multi-stream initiative one task at a time is slow and error-prone: scopes overlap, work streams get dropped, and cross-task dependencies go unrecorded. This Skill decomposes a complex brief into N independent sub-tasks and initializes them in parallel with validated, disjoint scopes. ## Core Features & Use Cases - Orchestrator-worker decomposition: An Opus-tier orchestrator validates the decomposition (unique slugs, disjoint scopes, DAG cross-links, full deliverable coverage) before dispatching Sonnet workers that each run a full task-init. - Parallel task folder creation: Each worker writes the 5 mandatory files (README.md, TASK_STATE.md, SOURCE_OF_TRUTH.md, DECISIONS.md, IMPLEMENTATION_PLAN.md) into its own task folder under projects/<client>__<project>/active/. - Cross-task index and audit trail: The orchestrator merges results into INITIATIVE_INDEX.md with typed cross-link relations, runs an orphan scan, and appends tamper-evident entries to VERIFICATION_LOG.jsonl. - Use Case: Migrating from REST to GraphQL across api-server, mobile-app, and web-app: the Skill splits the brief into three sub-tasks, initializes all three task folders in parallel, and records their shared-contract cross-links in the initiative index. ## Quick Start Ask the agent to run task-init-fleet on your multi-stream brief, for example: "Use task-init-fleet to decompose this REST-to-GraphQL migration brief across api-server, mobile-app, and web-app into parallel sub-tasks."

Frequently Asked Questions about task-init-fleet

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

FAQPage Schema
How do I initialize multiple tasks in parallel from one brief?▼

Run task-init-fleet with your full brief and a client__project identifier. The orchestrator decomposes the brief into sub-tasks, validates that scopes are disjoint, then dispatches parallel workers that each create a task folder with the 5 mandatory files.

When should I use task-init-fleet instead of task-init?▼

Use task-init-fleet only when the brief contains 3 or more logically independent work streams that can run in parallel. For one cohesive task, sequential dependencies, or unclear sub-task scopes, use task-init or run decision-interview on the parent brief first.

What happens if the decomposition has overlapping scopes?▼

The orchestrator validates uniqueness, scope disjointness, DAG cross-links, complexity tiers, and deliverable coverage before dispatch. If any check fails, the run stops with a NO_OP_TRACE listing the failures and routes to decision-interview instead of creating folders.

Can task-init-fleet mix multiple client projects in one run?▼

No. Mixing client_project values across sub-tasks in one fleet run is forbidden because INITIATIVE_INDEX.md writes must stay scoped to a single project. Re-run the command once per client_project.

What is the maximum number of parallel workers?▼

The default max_fanout is 10 workers, with an absolute ceiling of 20 to keep parallel project-memory access manageable. The barrier convergence waits for all workers or a 10-minute timeout, and partial results are allowed.