convez.ai
SkillMarketing & VendasIntermediário
hkaur51

Cold DM Skill para LinkedIn

hkaur51/ColdDMs

Skill de outreach LinkedIn com fórmula testada e 30% de conversão

Ver no GitHub
InicialPluginScore · de 0 a 100

Skill reutilizável para redigir mensagens diretas de prospecção no LinkedIn que geram respostas e coffee chats. Inclui fórmula de três sentenças, guia de pesquisa do prospect, sequência de follow-ups e todo o playbook de outreach validado com 30% de taxa de conversão.

Instalação

$npx convez-ai hkaur51/ColdDMs

1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.

Métricas

Estrelas
0
Forks
1
Acompanhando
0
Saúde
28%
Último commit
há 5 meses
Versão
—
Issues abertas
0
Contribuidores
1

Casos de uso

  • Redigir cold DMs personalizados no LinkedIn com gancho específico
  • Pesquisar e analisar perfil de prospect antes de abordar
  • Sequenciar mensagens de follow-up e thank-you notes

Veredito

Skill prático e bem documentado, baseado em experiência real (30% conversão). A força está no rigor metodológico de pesquisa e na fórmula testada, não em automação mágica. Recomendado para quem quer estruturar prospecção B2B no LinkedIn com seriedade.

README

A tradução para o português deste README ainda está na fila.

Cold DM Skill

A reusable skill for writing cold and warm outreach DMs that actually get replies, plus the thank-you notes, follow-ups, and post-comments around them. The full playbook lives in SKILL.md. This README covers what's here, how to use it, and the part that makes or breaks every message: how to research a person and read their work before you write a word.

This guide is the result of cold DM-ing people on LinkedIn and reiterating the content and outreach messages over four months. It produced a ~30% conversion rate from a cold DM to coffee chat. Folks who I had genuine interest and experience overlap with replied 90% of the time.

What's in here

  • SKILL.md — the playbook: the three-sentence formula, what works, what to cut, accuracy rules, the editing loop, recipient calibration, and sequencing.
  • README.md — this file. explicitly explains how to research the person you are sending a cold dm to. the researching is the most important part of this process. I spent an average of 10-15 mins looking into the person's background. if their work did not align with my past experience or future ambitions, i dropped the person.

How to use it

As a Claude skill: keep the folder structure (cold-dm/SKILL.md) and install or load it. The description in the frontmatter triggers it whenever you ask for outreach help or paste a profile.

As a plain reference: open SKILL.md and follow it by hand. Paste a target's profile and your own background into any chat and ask for a draft against the formula.

Either way, the workflow is:

  1. Research the person (see below).
  2. Draft against the three-sentence formula in SKILL.md.
  3. Run the cutting passes until every sentence earns its place.
  4. Send the small ask (a 15-minute chat). Escalate later.

How to research a person and read their work

A cold DM only works if the hook is specific, real, and recent. That comes from research, not guessing. Generic praise ("I admire your work") fails because anyone could write it. A line that quotes their actual finding from a paper they wrote years ago can't be faked, and that's exactly why it lands.

Budget more time on research than on writing. The message is short; the homework is the job.

Step 1: Read the whole profile, not just the headline

Go past the current title. Pull from:

  • Their "about" / bio — the throughline they chose to tell about themselves.
  • Full work history — the order of moves matters. A pivot, a return, a jump between fields, time spent founding something. The shape of a career is often the best hook.
  • Education and side details — a thesis topic, a fellowship, a patent, a sport, a second language. Specific and rarely mentioned by others.
  • Volunteering / causes — often where the genuine mission lives.

Write down the 2-3 most specific, distinctive, real details. These are your hook candidates.

Step 2: Read what they've actually published or posted

This is the highest-value research and the part most people skip.

  • Recent posts — what are they talking about right now? React to a current idea, not a stale one.
  • Articles, papers, reports, whitepapers — find the actual document and read it. Pull a specific finding, statistic, or argument. Quoting a real result ("X% prefer A over B") is far stronger than summarizing vibes.
  • Older work — dig back years. Referencing something from early in their career signals effort almost nobody makes, and flatters without grovelling.
  • Talks, podcasts, panels — sometimes the most candid, quotable material.

When you read their work, look for one of two things:

  1. A finding striking enough to react to honestly.
  2. A problem or gap in their work that overlaps with a gap you've hit in yours. That intellectual overlap is the strongest hook of all: same problem, two vantage points.

Where to actually find their work (search tactics)

Finding the material is its own skill. What works, roughly in order:

  • Comb their LinkedIn posts AND reposts. The Activity tab (Posts, Comments, Documents) shows what they care about right now. Reposts are a tell too: what they amplify says as much as what they write. Don't forget the "Featured" section, where people pin their most important work.
  • Search their name + their firm or school. A plain web search like [Full Name] [Firm or School] surfaces bios, articles, interviews, and announcements that never appear on their profile.
  • Search for published work and talks specifically. Add a format or content word: [Full Name] [Firm] PDF, or swap in talk, podcast, interview, panel, webinar, keynote. PDFs are often papers, whitepapers, or slide decks with quotable specifics.
  • Use search operators to narrow. "Full Name" in quotes forces an exact match. [Full Name] filetype:pdf finds documents. site:firm.com [Full Name] finds their employer's own bio/author page (consulting firms, universities, and labs all maintain these, often with a fuller history than LinkedIn).
  • Go to the source databases for researchers. Google Scholar, ResearchGate, ORCID, and PubMed surface papers, co-authors, and citation trails. A faculty or staff profile page is often the richest single source.
  • Find spoken material. Search YouTube and podcast platforms by name; conference agendas and speaker pages list talks (and affiliations). Spoken content is often the most candid and quotable.
  • Find where they're quoted. News articles, press releases, and company blog posts that quote them give you current, attributable lines to react to.
  • Check personal channels. A Substack, Medium, personal site, or X/Twitter account is sometimes where the realest thinking lives.

How to find their email

  • The cleanest source: their own published work. Papers and reports almost always list a corresponding author or a point-of-contact email. Whitepapers and articles frequently print the author's work email at the end.
  • Deduce the firm's email pattern. Once you have one known address from a paper, you know the firm's format (e.g., first_last@, first.last@, flast@). Apply that pattern to anyone else at the same firm.
  • Faculty/staff pages and conference listings often print emails directly.
  • A warm intro still beats a cold email. If a mutual contact can introduce you, take that over any address you dig up, especially for recruiters.

Step 3: Find the genuine connection

The hook has to tie to the sender, not just flatter the recipient. Look for the real overlap:

  • Same field, problem, or mission.
  • A path that rhymes with the sender's (a similar pivot, a shared background, a parallel move).
  • A finding in their work that maps onto the sender's own experience.

If there's no real connection, this may be the wrong person to message. A great message to the right person beats ten generic ones.

Step 4: Verify everything before it goes in

Research is worthless if the detail is wrong, and a single error undercuts the whole message. Before writing:

  • Check the verb. Did they found, lead, or join the thing? They're different claims.
  • Check the story. Does the timeline actually support the arc you're about to describe (a return, a pivot, a sequence)?
  • Check identity. If you're referencing a co-author or a name, confirm it's the same person and not someone with a matching first name.
  • Check the specifics. Country, dataset, dates, statistics. If you cite a published finding, cite it exactly.

If a detail can't be verified, leave it out. A vaguer true message beats a specific false one.

Step 5: Match the research to the recipient

How deep to go, and what to surface, depends on who they are:

  • Senior leaders / partners — find their published work or a signature idea; build one sharp question around it.
  • Recruiters — research the hiring context (office, role, timing) more than personal detail; ask a tactical question.
  • Peers / mid-level — the career-path or shared-background hook usually works; recent posts and the shape of their moves are enough.

The core idea

The research is what separates a message that gets ignored from one that gets answered. The formula in SKILL.md is simple on purpose. The work is reading carefully enough to say something only this person could receive, and only this sender could send.

Ver o original no GitHub

Bom para

Relacionados