Claude Code Session Handoff
Write a verified handoff so a fresh session, subagent, or teammate can continue coding work cold: state, dead ends, next step.
Before & after
Same prompt, same agent — the only difference is the skill.
Handing off work to someone else is important for team continuity. You'll want to make sure they understand what you've done, what's still in progress, and what decisions you made along the way. Writing down your progress and any blockers is helpful. You should also share any relevant context—like what the customer asked for, what you tried, and why you made certain choices. It's good to have a conversation with the person taking it over so they can ask questions. The goal is to avoid any surprises when they start working on it.
WORK HANDOFF PACKAGE STATUS SUMMARY • Task: Feature X (due 2026-08-15) • Completed: 60% (backend logic done; UI incomplete) • Blockers: API endpoint spec changed 2026-07-29; frontend dev paused • Handoff to: teammate name, timezone UTC+1 DELIVERABLES CHECKLIST ✓ PR #347: backend logic + tests (code review passed) ✓ Spike doc: API changes breakdown (3 endpoints affected) ✗ UI component: 40% done (branch: feature/ui-work) ✗ Integration tests: blocked on API stability CONTEXT & DECISIONS DECISION 1: Use caching layer for API responses → Why: 2.3s latency before; 200ms after cache (90% improvement) → Risk: cache invalidation edge case (documented in code comments) DECISION 2: Split into 2 smaller PRs instead of 1 large → Why: easier review; independent deploy possible → Impact: teammate can land backend solo if UI slips NEXT STEPS (Priority order) 1. Wait for API v3.2 release (ETA 2026-08-02) 2. Update frontend schema; re-run integration test suite 3. Build UI in 2 sprints: form input (3 days), validation UI (2 days) 4. Code review + deploy together in 1 batch QUESTIONS TO DISCUSS • Is API v3.2 release timeline still 2026-08-02? • Should we include validation error UX in scope, or defer to v2?
About this skill
Running out of context or ending the day mid-task? This skill writes the handoff note that lets a new Claude Code session, a subagent, or a teammate continue in minutes, and it separates verified from remembered so nothing is marked done without evidence.
What it does
- Marks DONE only with evidence from the session (test output, commit SHA), and gives the reader exact re-verify commands (or runs them when you ask)
- Separates verified DONE from IN PROGRESS and NOT STARTED, tied to files and commit SHAs
- Records decisions with reasons and the dead ends not to retry
- Writes self-contained subagent briefs with scope, context, and return format
- Keeps secrets out and uses absolute dates
What you get
A short note for quick breaks, or a full ready-to-paste handoff (saved to HANDOFF.md when you ask) with goal, verified state, decisions, dead ends, next step, re-verify commands, and open questions.
Best for
Claude Code users working on multi-session tasks, delegating to subagents, or passing work between people.
SKILL.md preview
---
name: session-handoff
description: Write a handoff note so a fresh Claude Code session, a subagent, or a teammate can pick up in-progress coding work cold - goal, verified state tied to files and commits, decisions with reasons, dead ends, exact next step, and how to re-verify - with nothing claimed as done that was not checked. Use when the user says "write a handoff", "I'm running out of context", "save state before I clear", "continue this in a new session", "brief a subagent on this", "hand this to another agent", or "wrap up so I can pick this up tomorrow".
---
# Session Handoff
… (sign up to view the full skill)More ai & agents skills
View all AI & Agents skills →New Skill Scaffolder
Create a new Claude Code skill correctly: right mechanism, valid name and folder, a triggering description, and a smoke test.
Skill Library Lint and Overlap Report
Lint every skill in a repo at once: frontmatter, name-folder match, weak descriptions, overlapping triggers, broken refs, plus a CI script.
Self-Verifying Agent Loop: Plan, Check, Retry, Close
Turns a goal into a task plan with machine-run checks, executes with retry caps, escalates, and refuses to close until verified.
Skill Test Runner (Trigger and A/B)
Test a skill in fresh sessions: ready-to-run prompts, commands, and rubric for trigger recall, false triggers, and blind A/B.