# Task Bootstrap

> Use when beginning any plan, audit, investigation, modification, review, test, or expensive tool action before the primary objective and completion conditions are explicit.

- Skill: `ihabkhaled/task-bootstrap` (Agent Skill)
- Install (CLI): `npx skillmds@latest add ihabkhaled/task-bootstrap`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ihabkhaled/task-bootstrap/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Security
- Author: ihabkhaled (https://skillmd.com/u/ihabkhaled)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/ihabkhaled/task-bootstrap

---


# Task Bootstrap

## Core principle

Executive control loads before planning; planning without a locked objective can become the first overthinking loop.

## Procedure

1. Read applicable platform and repository instructions.
2. State the exact primary objective and observable success condition.
3. Define the minimum finite Definition of Done, included scope, and excluded scope.
4. Name the current blocker or none, one active work item, and the next deliverable.
5. Route only task-relevant context and inspect the smallest evidence needed.
6. Create a bounded executable plan and begin the first deliverable.

## Required output

Objective, success, DoD, scope, out-of-scope, blocker, active item, nesting depth, retry count, verification state, and next deliverable.

## Limits

WIP 1; start at nesting depth 0; no broad audit unless required by the objective.

## Common mistakes

Do not write a giant plan before bootstrap or convert implementation into a repository-wide audit.

## Stop condition

Stop bootstrap when the agent can name exactly what done means and the first evidence-producing action.