task-init

Creates the task folder and base memory files for a new engineering task.

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? AI-assisted engineering work loses context between sessions because decisions and plans live only in chat history. This Skill initializes a persistent on-disk task folder so every new task starts with structured, resumable state instead of a blank conversation. ## Core Features & Use Cases - Task folder scaffolding: Creates the dated task directory under projects/<client>__<project>/active/ with README.md, TASK_STATE.md, SOURCE_OF_TRUTH.md, DECISIONS.md, and IMPLEMENTATION_PLAN.md. - Context seeding: Pulls existing project memory (PROJECT_CHARTER.md, REFERENCES.md, prior LEARNINGS.md) and optionally a vetted issue-tracker MCP item into the new task files. - Pipeline recommendation: Assesses task complexity and recommends an Express, Standard, Disciplined, or Strict command pipeline, plus multi-repo and worktree-isolation handling. - Use Case: Starting a new feature, you describe the objective and client/project; the Skill creates the full task memory structure with a tamper-evident verification log so later sessions can resume exactly where you left off. ## Quick Start Initialize a new task for client acme project web-app with the objective to add OAuth login support.

Frequently Asked Questions about task-init

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

FAQPage Schema
How do I start a new AI-assisted engineering task with persistent state?▼

Run task-init with the task objective, client and project identifiers, and a task slug. It creates the dated task folder with README, TASK_STATE, SOURCE_OF_TRUTH, DECISIONS, and IMPLEMENTATION_PLAN files so context survives across sessions.

When should I use task-init versus resume-from-state?▼

Use task-init only when starting a task from zero with no existing task folder. If a task folder with TASK_STATE.md already exists, use resume-from-state to continue; for progress updates on an active task, use sync-task-state instead.

Can task-init seed a task from an issue tracker ticket?▼

Yes, through a gated opt-in path. When a vetted, human-approved issue-tracker MCP is connected, task-init pulls the item's title, body, identifier, and URL to seed the task description and provenance; on failure it asks you to paste the item manually.

Does task-init support tasks spanning multiple repositories?▼

Yes. When two or more product repositories are provided with identifier, path, base branch, and role tag, it emits a Repositories section in SOURCE_OF_TRUTH.md. Single-repo tasks keep the standard active codebase field unchanged.

What are the limitations of task-init?▼

It never writes production code, never provisions git worktrees (that lives in task-workspace), and never auto-loads the human knowledge folder. It also stops and asks rather than creating files when the workflow spec or task repository cannot be resolved.