Session Handoff logo

Session Handoff

Community
Innei
session-handoff

Produce a self-contained handoff prompt for another agent (Codex, a fresh Claude session, a teammate) when the user wants to delegate continued work. Triggers on: "写一个 prompt 给 codex"、"交接一下"、"让 xxx 跟进"、"summary 一下再写个 prompt"、"handoff to another agent"、"write a prompt so X can continue", or any request to capture the current session state for continuation elsewhere.

Overview

PublisherInnei
RepositorySKILL
Skill namesession-handoff
Stars
81
Forks
2
Bundled files
Instructions only
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by Innei on GitHub. Read the source before you install it.

Installation

Install the Session Handoff AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/Innei/SKILL.git /tmp/SKILL
mkdir -p .claude/skills
cp -r /tmp/SKILL/skills/automation/session-handoff .claude/skills/session-handoff
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Session Handoff in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Session Handoff on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Session Handoff is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

session-handoff

Write a handoff prompt that lets a cold receiving agent continue the current work without re-asking the user for context. The receiving agent has zero knowledge of this session — the prompt must be self-contained.

When to invoke

  • User explicitly asks to hand off: "让 codex 跟进", "write a prompt for X", "交接给 xxx".
  • User asks for a session summary that will be forwarded to another agent.
  • User wants to pause and capture state for later continuation.

Do not invoke when the user just wants a status update or a summary for themselves — those are conversational, this produces a delegation artifact.

Required sections

The output is a single Markdown document the user will copy and paste. It must contain the following sections, in order, with the section titles as written (Chinese / English mix is fine — match whatever the rest of the session is using):

  1. 用户与环境 / User & Environment

    • Who the user is (name, handle, role) and contact if known.
    • Working directory, current branch, target merge branch.
    • Local services / databases (host, port, db name, credentials) the receiving agent will need. Redact .env contents and cloud tokens.
    • Dev workflow commands actually in use (bun run dev:spa, pnpm xxx, etc.).
    • User preferences and hard constraints (git rules, commit conventions, "do not touch" list, lint/test scoping rules, output style).
  2. 当前正在处理的事项 / Current Task

    • Top-level goal in 1–2 sentences.
    • The current blocker or open question.
    • If a previous approach was abandoned, say why so the receiver doesn't retry it.
  3. 本会话已完成的工作 / History

    • Chronological or grouped bullets.
    • Each bullet names: what changed, which files/lines, the rationale, and any user feedback that shaped the decision.
    • Include deprecations/migrations touched, not just additions.
  4. 下一步待做 / Next Steps

    • Numbered, concrete actions.
    • For each: files to touch, the specific change, and any preconditions.
    • Inline open questions that the receiver must ask the user before committing.
    • State which path you recommend and why, so the receiver can challenge the recommendation instead of rediscovering it.
  5. 关键文件速查 / Key Files Index

    • Grouped by concern (UI / services / server / schema / packages / tests).
    • One line per file explaining its role in this task.
  6. 调试 / 验证方法 / Debug & Verification

    • Exact shell commands to verify state (DB queries, log filters, curl invocations).
    • The test scenarios that should pass before shipping.
    • How to reproduce the original bug (if applicable).
  7. 需要用户定夺的点 / Open Decisions

    • Policy or UX questions the receiving agent must NOT silently choose.
    • Phrase each as a question with the options already identified.

Writing rules

  • Self-contained. No phrases like "as we discussed earlier" or "per the previous exchange". Inline the context.
  • Concrete over abstract. "改 foo.ts:42MAX_RETRIES 从 3 到 5" beats "tune retry behavior".
  • Backticks around file paths, commands, identifiers, and DB rows.
  • Line numbers whenever referencing a specific location — paths alone rot less, but line numbers make the handoff immediately actionable.
  • Shell blocks for DB queries, gh commands, bunx invocations. Use real, pasteable snippets; don't pseudocode.
  • Quote user feedback verbatim when it shaped a decision — the receiver needs to know what the user said literally, not a paraphrase.
  • No filler. Skip "this was a great discussion", emoji, or apology prose.
  • Redact secrets. .env, tokens, API keys never go into the prompt. If the receiver needs them, tell them to ask the user.

Presentation to the user

  • Do not emit the handoff prompt inline in chat. Write it to a file instead, then open that file for the user.
  • Steps, in order:
    1. Generate a random filename under the system temp directory, e.g. /tmp/session-handoff-<random>.md (use a short unique token — timestamp, uuidgen | head -c 8, $RANDOM, etc.). Avoid collisions with existing files.
    2. Use the Write file tool to write the full handoff prompt (Markdown, all required sections) to that path. Do not wrap the content in an outer fenced code block — the file is the artifact.
    3. Run open <path> (macOS) via the shell tool to open the file in the user's default Markdown viewer/editor.
    4. Reply with exactly OK — no preface, no summary, no path echo, no fenced block of the content. The file + open call is the delivery.
  • Do not execute the handoff yourself (don't spawn the other agent). Your job is to produce the artifact file.

Quality bar

Before emitting, check:

  • Could a cold teammate start from this prompt and work for an hour without needing to ask the user anything other than the listed Open Decisions?
  • Are all file paths real? No invented paths.
  • Are user-visible decisions and trade-offs preserved, not averaged out?
  • Are there concrete shell commands in the debug section?
  • Did any secret sneak in? Scan for tokens, passwords, DSNs.

If any answer is "no", revise before showing to the user.

Frequently asked questions

What does the Session Handoff AI skill do?

Produce a self-contained handoff prompt for another agent (Codex, a fresh Claude session, a teammate) when the user wants to delegate continued work. Triggers on: "写一个 prompt 给 codex"、"交接一下"、"让 xxx 跟进"、"summary 一下再写个 prompt"、"handoff to another agent"、"write a prompt so X can continue", or any request to capture the current session state for continuation elsewhere.

Why use Session Handoff on TypingMind?

Because you install it once and use it with any model. Session Handoff is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Session Handoff in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/Innei/SKILL/tree/main/skills/automation/session-handoff. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Session Handoff?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Session Handoff?

As many as you like. As long as a model supports skills, you can use Session Handoff with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Session Handoff AI skill free?

It is published on GitHub by Innei. Check the repository for licensing terms. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇