Baton logo

Baton

Community
blader
baton

Use when handing off an in-progress session to another agent or a future session — ending a session with unfinished work, "pick up where I left off", passing a task to a fresh agent/another window/codex, running low on context, or persisting intent + learnings before context is lost. Covers both dropping a baton (writing the handoff) and grabbing one (resuming from it).

Overview

Publisherblader
Repositorybaton
Skill namebaton
Stars
62
Forks
3
Bundled files
Instructions only
LicenseMIT
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 blader on GitHub. Read the source before you install it.

Installation

Install the Baton 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/blader/baton/tree/main
  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/blader/baton.git \
  .claude/skills/baton
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Baton 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 Baton 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 Baton 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.

Baton

Overview

A relay only works if the baton is passed cleanly. When you hand work to another agent — a fresh window, a future session, a different model — that runner starts with zero of your context. A baton is a single markdown file that lets them grab the work and sprint, instead of re-deriving everything you already know.

Core principle: capture what the next runner can't reconstruct from the code or git history — your intent, the dead ends you already ruled out, and the exact next step — grounded in verified repo state, not your chat narrative.

Two sides to this skill: Drop a baton (you're handing off) and Grab a baton (you're picking one up).

When to Use

Drop a baton when:

  • A session ends with work unfinished and someone/something will continue it.
  • The user says "hand this off", "pick up later", "pass this to another agent".
  • You're running low on context and want to checkpoint before it's summarized away.
  • You're about to dispatch the rest of the work to a separate agent/session.

Grab a baton when:

  • Starting work in a repo where a prior session may have left one (check first).

Don't use for: a finished task (no handoff needed), or durable architecture rationale (that belongs in a long-lived design doc, not a baton).

Where the Baton Lives

Default destination: .baton/ at the repo root — a dedicated folder for handoffs, created if it doesn't exist. It works in any repo, so there's no other location to choose.

.baton/ is local scratch, not committed. The first time you drop a baton, ensure it's git-ignored — if .gitignore doesn't already cover it, add a .baton/ line.

Filename: <YYYY-MM-DD>-<short-slug>.md (e.g. .baton/2026-06-13-formula-eval-rounding.md).

The next agent finds the latest with ls -t .baton/.

The Iron Rule: Verify Before You Write

A baton built from the chat story instead of repo reality is worse than none — it sends the next runner the wrong way with false confidence. Before writing, run git status and git branch --show-current and confirm what is actually committed vs. uncommitted. The "State of play" section must reflect runtime truth. If reality contradicts the narrative, the baton documents reality and flags the gap.

Baton Format

Fill every section. If one is genuinely empty, write "none" — don't delete it (the next runner relies on the shape being consistent).

markdown
# Baton: <one-line title of the work>

**TL;DR:** <1–3 sentences: what this is, where it stands, the single next action.>

## Intent & Goal
Why this work exists and the desired end state. Include the acceptance
criteria — ideally the exact command/test that must go green.

## State of Play (verified against git)
- Branch: <branch> — confirmed via `git branch --show-current`
- Committed: <what's landed>
- Uncommitted / in flight: <what's in the working tree, from `git status`>
- Verified working: <what you've actually run and confirmed>
- Assumed / unverified: <what you believe but haven't proven>

## Learnings & Landmines
The highest-value section — what the next runner can't get from the code:
- Non-obvious discoveries.
- **Dead ends already ruled out** (so they don't repeat them) and *why*.
- **Do NOT touch** zones and the reason.

## Pointers
- Key anchors: `path/to/file.ts:123` — what's there.
- Relevant docs / prior batons / PRs.
- Commands to get oriented or reproduce: `<cmd>`.

## Next Steps
Ordered, concrete, immediate action first. Not "finish the feature" — the
literal next move.

## Open Questions
Unresolved decisions + your current lean (so they can proceed if no one answers).

Grabbing a Baton (resuming)

  1. Find it: ls -t .baton/ — read the newest.
  2. Trust but verify: re-check "State of play" against real git status before acting; the repo may have moved since the baton was dropped.
  3. Continue from Next Steps.
  4. Retire it when the work lands: delete it if it was only scratch, or fold durable learnings into a canonical doc and then delete. Don't let stale batons pile up in .baton/.

Common Mistakes

MistakeFix
Baton restates the diffThe diff is already in git. Capture intent, dead ends, next step.
Built from chat, not git statusVerify first — a confidently-wrong baton is worse than none.
Vague next step ("continue the work")Write the literal next action.
Dead ends omittedNaming what didn't work is the highest-value content — it saves a full re-derivation.
Scattered handoff locationAlways .baton/ at repo root — never elsewhere, or the next agent can't find it.
Stale batons pile upRetire it when the work lands: delete, or fold learnings into a canonical doc.

Response Behavior

After dropping a baton, state the exact path written and a one-line summary of what's in it.

Frequently asked questions

What does the Baton AI skill do?

Use when handing off an in-progress session to another agent or a future session — ending a session with unfinished work, "pick up where I left off", passing a task to a fresh agent/another window/codex, running low on context, or persisting intent + learnings before context is lost. Covers both dropping a baton (writing the handoff) and grabbing one (resuming from it).

Why use Baton on TypingMind?

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

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

Which AI models can use Baton?

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 Baton?

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

Is the Baton AI skill free?

Yes. It is published on GitHub by blader under the MIT license. 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 👇