Linkedin Unibox Reply logo

Linkedin Unibox Reply

Community
timscheuerai
linkedin-unibox-reply

Draft LinkedIn DM replies in the author's voice for Unibox triage and 1:1 follow-ups, one at a time, and never send without the author's yes. Use to triage interested or unanswered LinkedIn threads, draft a DM reply, or rewrite a generic suggested reply into the author's voice.

Overview

Publishertimscheuerai
Repositorylinkedin-funnel
Skill namelinkedin-unibox-reply
Stars
60
Forks
20
Bundled files
1
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.

  • 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 timscheuerai on GitHub. Read the source before you install it.

Installation

Install the Linkedin Unibox Reply 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/timscheuerai/linkedin-funnel/tree/main/skills/linkedin-unibox-reply
  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/timscheuerai/linkedin-funnel.git /tmp/linkedin-funnel
mkdir -p .claude/skills
cp -r /tmp/linkedin-funnel/skills/linkedin-unibox-reply .claude/skills/linkedin-unibox-reply
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Linkedin Unibox Reply 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 Linkedin Unibox Reply 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 Linkedin Unibox Reply 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.

Workspace: all context paths below (identity, audience, strategy, voice, brand, raw, output, index, log, workspace preferences and context scripts) resolve from context/ in the kit root. Read both the root AGENTS.md and context/AGENTS.md. Skill-local references/ resolve from this skill directory.

LinkedIn Unibox reply

Draft 1 LinkedIn DM reply per thread in the author's voice, aimed at the smallest useful next step: help first, then signup, activation or, for qualified buyers, a call. Show the thread link. Never send unless the author says yes to that exact draft. Thread reading and sending run through oxygen-unibox; the conversation framework is guides/dm-conversion.md.

Voice sources: voice/linkedin-voice.md, compressed for DMs, and the author's own past messages. Avoid the shapes in references/anti-patterns.md.

DM, comment, post

SurfaceLengthJob
LinkedIn DMShort: usually under 240 characters, 1 to 4 short paragraphsAnswer their last message. Specific, founder to founder
Public comment1 to 3 linesA real observation or question; almost never a CTA
Post or articleLong formlinkedin-copywriter

Voice rules

  1. Answer the thread. Reference their last question, constraint or stack. No product monologue: in a live buyer thread, react and ask about the next piece of their setup; explain the product only when they ask.
  2. Short and specific. Everyday words with precise nouns. Keep the author's informal rhythm without a slang quota. Links on their own line, no sign-off.
  3. Ship the artifact. When they asked for the repo, recording or link, send it in the same message.
  4. Peer tone. Credit their work; compare tools fairly.
  5. Admit uncertainty when true. Never invent demos, timelines, capabilities, prices, discount codes or the author's plans. Where 1 is needed, write [AUTHOR: ...] and hold the item.
  6. No calendar push by default. Call links only for qualified buyers or when they asked (see "Route"). Chit-chat and networking get an answer in the thread, not a call.
  7. 1 clear next step at most: a link, a question or a deliverable. The closing question builds on what they said, not a stock question.
  8. Match their language and register.
  9. Never copy a tool's suggested reply. Rewrite from the thread in the author's words.
  10. Late replies: when their last message is more than 3 days old, own the delay in a few words, with the real reason when there is 1. Vary the wording across a batch; the same apology on every draft reads like a bot.
  11. No em dashes. A period, comma or "and".

Establish who still needs an answer

Triage status and "needs reply" filters are candidates, not proof. Reread each thread and exclude it when the latest message is outbound or the order of messages is unclear. Read enough history to see earlier links, questions, signups and meeting commitments. Before treating an outbound message as the author's own voice, check it was not sent by a Sequence or another automation. If the person also talks to the author in Slack or email, check there before drafting.

Route

ThreadDraft
Resource requestThe link plus 1 question about their work. 5 or more for the same resource: the DM-side delivery in lead-magnet-delivery
Compliment or new connectionShort thanks and what they are working on
Prospect, qualified (fits the buyer rubric from guides/qualification.md and shows a concrete need)Answer, then offer the call as a question
Prospect, unknown or weaker fitAnswer, then the self-serve start (the website or signup) with help offered if questions come up
Signed-up user or customerHelp with where they are stuck; onboarding calls only when qualified or paying
Message that needs no answer (a thanks, "done" after booking, "will check it out")No reply
Support problemThank them, ask where they are stuck; hold anything that needs a fix, refund or account change
Someone pitching the authorNo reply, unless the author asks
Event, podcast, partner, money, legal or personalHold for the author

Qualify before any call link: company fit, the contact's role, a concrete need and intent. A large employer or a senior title alone is not enough; unknown is not qualified. When someone says they will buy, help them take the next product step instead of adding a hurdle.

Workflow

  1. List interested and unanswered LinkedIn threads (see oxygen-unibox). A batch is bounded, not an inbox audit; say how much was covered.
  2. Read and route each thread as above.
  3. Draft 1 reply per thread and show a decision queue: ready to send (name, thread link, their latest point, fit evidence, route, the exact draft in a code block so line breaks show), then held items with the missing fact, then counts of already answered and noise. Buyers first, oldest unresolved first within each group.
  4. Gate: reread each thread right before presenting and again before sending. A new inbound or outbound message invalidates the old draft. Send only what the author approved, exactly; if the copy changes after approval, ask again.
  5. Send paced: approved replies go out a few minutes apart within the sender's limits. A send with an uncertain outcome is checked in the thread before any retry.
  6. Report: sent, skipped, failed and held. Track signups, first product actions and qualified bookings separately from links sent.
  7. Learn: when the author edits or rejects a draft, write the general rule behind it (never names or message text) into their voice context, and keep their final wording as a voice example.

Drafting on a schedule

A recurring drafting run can prepare the queue for review, for example every few hours. It drafts and presents; it never sends. Keep its state (last run, open drafts, verdicts, a private do-not-reply list) outside the repository, and run only 1 loop at a time: 2 loops on the same inbox draft the same threads and can answer 1 person twice. Advance the checkpoint only after every fetched thread was handled or kept for review. Approval stays with the author until they explicitly turn on sending for named, obvious cases; no streak or elapsed time switches it on.

Morning digest

Interested and unanswered threads, each as: name, 1-line last message, route, link. Then ask which to draft, or draft the queue if the author said "just start".

Paid profile or company reads to personalize a draft cost credits: quote the cost first and run them only on the author's yes.

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 Linkedin Unibox Reply AI skill do?

Draft LinkedIn DM replies in the author's voice for Unibox triage and 1:1 follow-ups, one at a time, and never send without the author's yes. Use to triage interested or unanswered LinkedIn threads, draft a DM reply, or rewrite a generic suggested reply into the author's voice.

Why use Linkedin Unibox Reply on TypingMind?

Because you install it once and use it with any model. Linkedin Unibox Reply 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 Linkedin Unibox Reply in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/timscheuerai/linkedin-funnel/tree/main/skills/linkedin-unibox-reply. 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 Linkedin Unibox Reply?

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 Linkedin Unibox Reply?

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

Is the Linkedin Unibox Reply AI skill free?

Yes. It is published on GitHub by timscheuerai 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 👇