# Handoff Task

> Use when work moves to another agent, session, person, or environment and the recipient needs enough state to continue without a sprawling narrative.

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

---


# Handoff Task

## Core principle

A handoff transfers objective, proof, state, blockers, and next action—not the entire reasoning history.

## Procedure

1. State objective and finite DoD.
2. List completed and remaining requirements.
3. Name blockers and important evidence.
4. List files touched and validation already run.
5. Record decisions and deferred findings.
6. Give one required next action and freshness warning if applicable.

## Required output

Objective, DoD, Completed, Remaining, Blockers, Evidence, Files touched, Deferred, and Required next action.

## Limits

Keep the handoff compact and factual.

## Common mistakes

Do not include raw chain-of-thought, repeated tool output, outdated hypotheses, or a new sprawling plan.

## Stop condition

Stop when a recipient can continue safely from one next action without rereading the whole repository.