A free Claude Code skill that builds your cold email strategy before you write a single line of copy, who to target, why they'd respond, and exactly how the sequence should run.
Cold Email Strategy is a Claude Code skill: a markdown playbook that makes your agent build the architecture behind a cold outbound sequence before any email gets written. It defines one coherent audience (segment, persona, problem, offer, trigger, proof, exclusions), writes a one-sentence reason that buyer would actually care right now, maps every touch in the sequence to a distinct purpose and a branch (what happens on a reply, an objection, a bounce, an out-of-office, a referral, an unsubscribe), sets sending capacity and domain-health rules, and designs one real experiment with a decision rule instead of vague "test stuff" advice.
The video's claim is that "this agent builds the real strategy first: who you're targeting, what makes them respond, and how the sequence should actually run." That held up in testing. Run against an invented business (a two-person AEO agency selling to Shopify store owners, made up for this test, not a real company), it produced a specific, non-generic target segment with real exclusions, a contact thesis grounded in a stated reason to care rather than a guess, a five-touch sequence with a distinct purpose and branch logic per touch, and a full reply, bounce, and governance routing table. All three parts of the claim held up concretely, and it correctly hands the actual email copy off to a separate skill rather than pretending to write your final messages for you, which is honest about its own scope.
One real thing to know before you run it: this skill designs the sequence, it does not go collect the evidence your outreach is based on. In the test, the strategy called for checking what ChatGPT and Perplexity currently say about a prospect's category, that's a real, useful trigger, but you'd need to actually go ask those AI assistants yourself (or use your own tool) for each prospect. The skill will tell you exactly what evidence it needs, it will not fetch it for you.
Any business owner, freelancer, or marketer about to run cold outbound who is tempted to jump straight into writing emails. This replaces the work of a strategist sitting down with you first and forcing you to answer who exactly you're contacting, why they'd care now, what happens after each touch, and when to stop, before a single sentence of copy gets written.
This copies the whole gtm-skills library onto your machine. You only need one folder out of it.
git clone https://github.com/LeadMagic/gtm-skills.gitThis drops just the cold-email-strategy skill into the folder Claude Code checks for project skills.
mkdir -p .claude/skills
cp -r gtm-skills/skills/outbound/cold-email-strategy .claude/skills/cold-email-strategyNote: the repo also documents two automated installers, scripts/cc-gtm.py and ./install.sh. In testing, cc-gtm.py crashed outright on the test machine's default Python (3.9) with a syntax error, it needs Python 3.12 or newer. ./install.sh depends on a gh skill command that a plain install of GitHub's CLI does not include, and when that command was missing it failed instead of falling back to a plain copy the way the docs describe. The cp -r above is the path that actually worked cleanly, and it lands the skill in the exact same place either installer targets.
Sample input used in testing: an invented two-person AEO agency, made up for this test, selling AEO audits and a monthly AI-citation monitoring retainer to US Shopify store owners doing $1-5M a year in revenue.
Abbreviated real output, following the skill's own output format:
Segment: US, non-Plus Shopify stores, $1-5M/yr, active paid-marketing
spend. Excludes Shopify Plus, dropship/private-label, B2B wholesale.
Contact thesis: rising paid-acquisition costs plus product-research
questions shifting to ChatGPT/Perplexity, where this brand doesn't
show up and a named competitor does.
Touch map: 5 touches over 15 days (email, LinkedIn connect, email,
email, email), each with one purpose and a branch for reply, no
reply, or objection.
Experiment: named-competitor AI quote vs. generic trend stat as the
opening line, with a stated sample size and a decision rule (adopt
only if one version shows at least 2x the reply rate).The skill's own bundled checker confirmed the output was complete and had no unfilled placeholders left in it.
scripts/cc-gtm.py (the "curated installer") crashes with a SyntaxError about an f-string and a backslash. That script needs Python 3.12 or newer. If you're on an older Python 3, use the cp -r copy method in Step 2 instead, it lands in the same place.
./install.sh fails with unknown command "skill" for "gh". That's GitHub's CLI missing the gh skill extension, not a problem with your setup. Skip it and use the cp -r copy method in Step 2, it's the reliable path.
The strategy calls for evidence you don't have yet (e.g. "check what AI assistants say about this prospect"). That's expected, the skill designs what evidence to gather, it doesn't gather it. Go check manually (ask ChatGPT or Perplexity the question yourself) or use your own research tool, then plug the answer back in.
You're not sure how many touches or what cadence to use. The skill deliberately avoids handing you a fixed universal number, it wants you to set timing from your own market, signal, and sending capacity. If you genuinely have no starting point, tell it that and ask it to propose a first-draft cadence you can adjust.
Built by LeadMagic (https://github.com/LeadMagic/gtm-skills), MIT license, 53 stars at test time. The video's claim held up exactly as stated: it produces a specific "who you're targeting, what makes them respond, how the sequence should run" strategy, not generic advice, and it stays honest about handing final copywriting off to a separate skill. Two corrections worth knowing: the repo's cc-gtm.py installer needs Python 3.12+, and install.sh's primary path can fail without falling back to a plain copy the way its own docs describe, use the manual copy method instead.
This is one agent. The 7-Day AI Native Challenge is how you learn to build the next ten yourself, live, with Dany.
Join the Challenge →Too busy to build it yourself? Focentra AI will build, install and train your team on it. Priced on the call.
Talk to Focentra AI