Handoff logo

Handoff

Organization
greenfield-inc
handoff

Prepare a verified resumption brief when moving work to another agent, device, or session, including local-to-cloud coding handoffs, and start the receiving agent with `runpane handoff`. Use for "hand this off," "continue on my Mac," "give this to Codex," or "save where we are."

Overview

Publishergreenfield-inc
RepositoryPane
Skill namehandoff
Stars
529
Forks
227
Bundled files
1
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.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by greenfield-inc on GitHub. Read the source before you install it.

Installation

Install the 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.
    https://github.com/greenfield-inc/Pane/tree/main/main/src/services/paneChatBundle/skills/handoff
  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/greenfield-inc/Pane.git /tmp/Pane
mkdir -p .claude/skills
cp -r /tmp/Pane/main/src/services/paneChatBundle/skills/handoff .claude/skills/handoff
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable 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 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 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.

Handoff

You are the outgoing collaborator. Give the next person or agent the verified context, decisions, and lessons needed to pick up where you left off.

Capture what matters

  • Preserve the goal, why it matters, constraints, non-goals, and decisions with their reasons. Distinguish approved work from suggestions.
  • Verify the current state. Record completed work, remaining work, meaningful failed attempts, blockers, and the next concrete action.
  • Link authoritative instructions, plans, evidence, and artifacts. Make the brief self-contained and focused on what the next worker needs.
  • Include pending approvals and the scope of authorized work. Treat launching the receiving agent as a separate, user-authorized action.

Make coding work portable

  • Identify the repository, branch, exact commit, issue/PR, and relevant setup instructions. Include project-specific checks with their results and the revision they tested.
  • Inspect staged, unstaged, untracked, and unpushed work. Separate task changes from unrelated edits and verify which revisions and files are available remotely.
  • Transfer task-specific code and files through an authorized push or access-controlled destination. Preserve unrelated work and identify any files still awaiting transfer.
  • Record required tools, services, and credential setup references. Keep secret values in their approved secret store and check the receiving agent’s plugin, skill, and network prerequisites.
  • Keep private code and patches in access-controlled locations. Identify the next step needed to resolve any transfer or permission blocker.

Write the note

Write the brief as the handoff note in references/note-template.md: a compaction of your session, in nine required ## sections, for a reader with none of your context.

Start the receiving agent

When the person asks for another agent to continue, on this machine or another of theirs, use runpane handoff. It works the same from any harness. "Hand this to Claude on my Mac" authorizes starting that agent; it does not authorize pushing.

  1. Find the machine's name: runpane workspace list lists the person's machines (a unique prefix such as macbook-pro works). Ask when more than one fits.
  2. Get the template outside the checkout, so it is never committed: runpane handoff --template > ~/handoff.md. Fill in every section; write "None" where one does not apply. Leave the front matter alone, and don't add instructions for the receiver: the CLI rewrites the front matter and appends them when it sends.
  3. Make the work reachable. Commit the task's files yourself; leave unrelated edits uncommitted. If the branch is not pushed, ask before pushing, then push it or pass --push (it commits every remaining change except the note as WIP and pushes; it never forces).
  4. Check the destination: runpane handoff "claude opus on <machine>" --note-file ~/handoff.md --dry-run. The destination is freeform: the agent (claude, codex, cursor), and optionally the model, the effort and the machine, such as "codex gpt-5 high on parsa-devbox" or "cursor on this machine". --machine, --agent, --model and --effort override the text. Effort is supported for Claude and Codex; Cursor effort is rejected before committing or sending.
  5. Send it: the same command without --dry-run. It finds the destination's saved repository with the same remote, writes the note to that machine's ~/.pane/handoffs/, and starts the agent. On another machine it gets a new Pane branched from yours; on this machine, run from the Pane that owns this checkout, it opens as a new tab in your Pane (1 feature = 1 worktree = 1 branch = 1 Pane), and the JSON's receiver.route and receiver.reason say which happened. If it fails, relay its message; it names the fix (Pane not running there, repository not saved there, branch not pushed).
  6. Tell the person where the receiver started (a new tab in your Pane, or a new Pane on which machine), and give them the agents status command it printed. The receiver pushes to your branch and reports back to your panel, so stop changing that branch.

The note is the handoff. Skip the two sections below unless the person also wants a saved page, a Grain workspace or a PR comment, or there is no receiving agent to start.

Save one authoritative brief

  • Honor an explicit destination. Otherwise, when Grain is connected, read its installed skill and update the existing task workspace; if none exists, create a clearly named one in Development Artifacts. Retain its ID for subsequent updates.
  • When the handoff lands in a Grain workspace and the session-trace skill is installed, attach this session's trace to it.
  • Make essential text directly readable to both humans and agents. Link existing artifacts and retain one authoritative brief.
  • For cross-device delivery, prefer a public-safe Grain share when public sharing is authorized by the request or an established user preference. Otherwise use an appropriately restricted destination or ask before publishing.
  • Inspect the full shared content for its intended audience. Use access-controlled GitHub when the brief requires private source or sensitive context. Return the verified share URL supplied by the service.
  • If Grain is unavailable or unsuitable, save the handoff as a page in the work's page bundle (see page) and open it for the person. With a posting grant, also update a clearly labeled handoff/status section on the existing PR, or the issue if there is no PR. Preserve the original intent, acceptance criteria, and other contributors' content; use a timestamped comment when editing the body would be disruptive.
  • If both destinations are unavailable, return a self-contained, copyable brief in chat with its save status. Create a new issue or PR when the user requests one.
  • Keep necessary local working files and verify that required material also has a remotely accessible copy. Report connected save failures and the outcome of any fallback.

Verify and deliver

  • Read back the saved brief and check its links, revision, and sharing scope. Verify public links in a signed-out context when possible, and state the verification status of recipient access to private resources.
  • Return the brief's link and a paste-ready instruction to resume, naming the first action and any transfer/access blockers.
  • Tell the receiving agent to read repository instructions and reconcile this checkpoint with the current branch, issue/PR, and artifact state before acting.

WSL handoff is currently unsupported and is rejected before committing or sending anything. Use a native repository. Remote shells must be Bash, Zsh, Sh, Fish, or PowerShell; cmd is unsupported. A staged note must be unstaged before using --push.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Handoff AI skill do?

Prepare a verified resumption brief when moving work to another agent, device, or session, including local-to-cloud coding handoffs, and start the receiving agent with `runpane handoff`. Use for "hand this off," "continue on my Mac," "give this to Codex," or "save where we are."

Why use Handoff on TypingMind?

Because you install it once and use it with any model. 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 Handoff in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/greenfield-inc/Pane/tree/main/main/src/services/paneChatBundle/skills/handoff. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use 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 Handoff?

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

Is the Handoff AI skill free?

It is published on GitHub by greenfield-inc. 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 👇