session-handoff
Use when approaching context limits (75-85%) to write comprehensive handoff documents for session continuity.
Skill by nocksock
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/nocksock-session-handoff
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: documents, writing
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/session-handoff
What's inside
- Skill name
session-handoff- Description
- Use when approaching context limits (75-85%) to write comprehensive handoff documents for session continuity. Also use when continuing from a handoff document.
- License
- MIT
- Instructions
- 185 lines in SKILL.md · view on GitHub
Files (2)
- SKILL.md
- handoff-template.md
SKILL.md (first lines)
# Session Handoff ## Overview Write comprehensive handoff documents when approaching context limits, and continue sessions from handoff documents. **Core principle:** The next session has ZERO context. Write for a stranger. **When to use:** - Context usage reaches 75-85% and work remains - User asks to pause and continue later - Continuing a session from a handoff document ## CRITICAL: Understanding Context Loss **The new session will have ZERO context from the current session.** This is not like "pausing" - this is a complete reset. The next agent: - Has no memory of this conversation - Does not know what files you looked at - Does not know what decisions you made or why - Does not know what the user said - Does not know what you learned about the codebase - Only knows what is written in the handoff document **If it's not in the handoff document, it does not exist for the next agent.** You are not "handing off to yourself" - you are writing instructions for a stranger who happens to have the same capabilities. Everything you currently "just know" from this session MUST be explicitly written down or it is lost forever. ## Monitoring Context Usage A hook injects context usage updates at 25%, 50%, 75%, then every 5% until 95%, then every 1% until 100%. You can also call the `checklimits` tool proactively. At each context update, assess: - How much work remains? - Is there risk of running out before completing the task? - Early updates (25%, 50%) usually don't have enough info to judge ## When to Wind Down **Target: Start winddown at 80-90% context usage.** - Not too early (waste context capacity) - Not too late (need context for winddown steps, and model performance degrades after 90-95%) **Trigger winddown when:** - Context usage is 75-85% AND - Remaining work likely exceeds remaining context AND - You can reach a reasonable stopping point ## Writing a Handoff Document ### Step 1: Finish Current Atomic Task Complete to a stopping point. Don't leave things half-done. ### Step 2: Write Handoff Document You MUST use the template in `./handoff-template.md`. **MUST NOT:** - Skip sections marked MUST - Write brief summaries assuming the next agent "will figure it out" - Use vague language like "continue implementing" **MUST:** - Fill every MUST section completely - Be specific about file paths, line numbers, exact state - Explain the "why" for every decision - Capture user preferences and quirks explicitly ### Step 3: Save to Predictable Location Save to: `docs/session-handoffs/YYYY-MM-DD-HH-MM-<topic>.md` Create the directory if it doesn't exist. ### Step 4: Ask User to Continue Use the question tool: > "I'm approaching my context limit. I've written a detailed handoff document ready to continue in a new session. Ready to continue?" > Options: Yes, continue in new session / Not yet, I have questions ### Step 5: Provide Continuation Prompt ``` I'm continuing a previous session. The handoff document is at: docs/session-handoffs/[filename].md Read that file and continue where the previous session left off. ``` The user will copy this into a new session. **For deep work sessions, use this prompt instead:** ``` I'm continuing a previous deep work session. Use the deep-work-session skill to continue. The handoff document is at: docs/session-handoffs/[filename].md Read that file and continue the deep work session autonomously. ``` ## Continuing From a Handoff Document When the user provides a handoff file path to continue from: ### CRITICAL: You Have No Prior Context **You are not "continuing" - you are starting fresh with a briefing document.** You have no memory of the previous session. You do not know: - What the user said before - What was tried and failed - What decisions were made or why - What the user's preferences are - What files are relevant The handoff document is your ONLY source of context. You MUST read it completely and trust its contents.
From nocksock/superpowers, MIT.
More from nocksock/superpowers
- 0.432score
deep-work-session
nocksock
Use when user explicitly mentions 'deep work' or signals substantial work ('let's build/implement X').
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
holistic-view
nocksock
Use this when given larger tasks, before starting the implementation, or during code reviews.
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
jotai-atom-architecture
nocksock
Use when designing Jotai state architecture, setting up atom composition patterns, managing large objects, or debugging re-render issues - ensures proper atom…
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
making-commits
nocksock
Use when ready to commit code changes.
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
component-authoring
nocksock
Use this when writing any kind of UI components in any language (heex, react, web components, and so on).
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
r3f-performance-optimization
nocksock
Use when optimizing React Three Fiber scenes, experiencing frame drops, high draw calls, or implementing 3D performance features - ensures systematic…
SkillDFreshness: D★ 1Claude Code +9 - 0.431score
code-cleanup
nocksock
Use when asked to clean up code, or after introducing a new larger feature.
SkillDFreshness: D★ 1Claude Code +9 - 0.423score
prioritise-todos
nocksock
Use this to rate and prioritise todos
SkillDFreshness: D★ 1Claude Code +9 - 0.423score
elixir-code
nocksock
ALWAYS use this skill, whenever you touch elixir code
SkillDFreshness: D★ 1Claude Code +9 - 0.423score
find-code-todos
nocksock
Use this to find todos in the source code
SkillDFreshness: D★ 1Claude Code +9 - 0.409score
brainstorming
nocksock
You MUST use this before any creative work - creating features, building components, adding functionality, or modifying behavior.
SkillDFreshness: D★ 1Claude Code +9 - 0.407score
finishing-a-development-branch
nocksock
Use when implementation is complete, all tests pass, and you need to decide how to integrate the work - guides completion of development work by presenting…
SkillDFreshness: D★ 1Claude Code +9
Signals
- Stars
- 1
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Feb 2, 2026 (8 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/nocksock/superpowers
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from nocksock
- 0.432score
making-commits
nocksock
Use when ready to commit code changes.
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
jotai-atom-architecture
nocksock
Use when designing Jotai state architecture, setting up atom composition patterns, managing large objects, or debugging re-render issues - ensures proper atom…
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
deep-work-session
nocksock
Use when user explicitly mentions 'deep work' or signals substantial work ('let's build/implement X').
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
holistic-view
nocksock
Use this when given larger tasks, before starting the implementation, or during code reviews.
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
component-authoring
nocksock
Use this when writing any kind of UI components in any language (heex, react, web components, and so on).
SkillDFreshness: D★ 1Claude Code +9 - 0.432score
r3f-performance-optimization
nocksock
Use when optimizing React Three Fiber scenes, experiencing frame drops, high draw calls, or implementing 3D performance features - ensures systematic…
SkillDFreshness: D★ 1Claude Code +9
Cite this page
session-handoff skill. Agent Mart. https://agentmart.dev/skills/nocksock-session-handoff (last updated 2026-10-24)