cold-email

Overview of cold-email skill

What the cold-email skill is for

The cold-email skill helps an agent write B2B outbound emails and follow-up sequences that sound human, concise, and commercially credible. It is built for sales outreach, prospecting, SDR workflows, and founder-led outbound where the goal is usually a reply, intro, or meeting request.

Best fit users

Use this cold-email skill if you already know the offer and audience, but want stronger messaging than a generic “write a sales email” prompt usually produces. It is especially useful for:

• SDRs and AEs writing outbound
• founders doing early sales outreach
• marketers supporting sales outreach
• operators creating repeatable cold-email usage patterns for teams

Real job-to-be-done

Most users do not need “an email.” They need a short message that:

• connects to a prospect-specific problem
• uses proof without sounding promotional
• asks for a low-friction next step
• extends into a sensible follow-up sequence

That is where this skill is more useful than a shallow template.

What makes this cold-email skill different

The repository is opinionated in ways that matter for output quality:

• it tells the agent to check for existing product marketing context first
• it emphasizes writing like a peer, not a vendor
• it uses frameworks instead of fixed templates
• it treats brevity as a performance constraint, not a style preference
• it includes references on subject lines, personalization, follow-ups, and benchmarks

In practice, that means the skill is better for message quality and reply-oriented outreach than for mass-template generation.

When this skill is not the right fit

This is not the best fit for:

• warm nurture or lifecycle email programs
• broader sales collateral
• compliance-heavy enterprise messaging where legal review dominates
• users who need deliverability setup, domain warming, or sending infrastructure guidance

The skill is about copy and sequence quality, not outbound ops.

How to Use cold-email skill

Install cold-email in your skills environment

Install from the repo with:

npx skills add https://github.com/coreyhaines31/marketingskills --skill cold-email

If your environment already supports local or synced skills, add it through that workflow, then confirm the skill appears under skills/cold-email.

Read these files first

For fast adoption, start with:

• skills/cold-email/SKILL.md
• skills/cold-email/references/frameworks.md
• skills/cold-email/references/personalization.md
• skills/cold-email/references/subject-lines.md
• skills/cold-email/references/follow-up-sequences.md
• skills/cold-email/evals/evals.json

This order matters: SKILL.md defines behavior, the reference files explain why, and evals/evals.json shows what good invocation looks like.

Check for product context before prompting

The skill explicitly expects the agent to look for:

• .agents/product-marketing-context.md
• .claude/product-marketing-context.md

If either file exists, use it before asking the user basic positioning questions. This reduces repetitive prompting and usually improves the first draft because the value proposition, proof points, and audience assumptions are already grounded.

Inputs the cold-email skill needs

The skill works best when you provide:

1. target audience or exact role
2. company type or segment
3. desired outcome: reply, intro, call, demo, referral
4. offer or value proposition
5. proof point: result, case study, customer, metric
6. research signals: hiring, funding, new initiative, tech stack, change event
7. any constraints: tone, word count, banned claims, CTA limits

Without these, the model can still draft copy, but it will default toward generic outbound language.

Turn a rough request into a strong prompt

Weak request:

• “Write a cold email for my SaaS.”

Stronger cold-email usage prompt:

• “Write 3 cold email variations for VPs of Marketing at mid-market B2B SaaS companies. We help teams measure which content drives pipeline. Proof: customers see 3x content-attributed revenue in 90 days. Use a peer-to-peer tone, keep each email under 90 words, give 3 subject lines per version, and end with a low-friction CTA. If product context files exist, use them first.”

Why this is better:

• gives segment, pain area, and proof
• sets brevity constraints
• requests multiple variants
• asks for the skill’s preferred CTA style

Use frameworks, not templates

A practical advantage of this cold-email skill is its framework library. The references include structures like:

• PAS
• BAB
• QVC

Use them intentionally:

• choose QVC for busy executives who need brevity
• choose PAS when the pain is obvious and expensive
• choose BAB when the transformation is easy to imagine

If you do not specify a framework, ask the agent to choose one and explain why it fits the prospect.

Ask for subject lines the skill is designed to write

The repository’s subject line guidance is unusually actionable: short, lowercase, internal-looking subject lines tend to perform better than polished marketing headlines.

Good request pattern:

• “Give me 5 subject lines, 2–4 words each, all lowercase, tied to the prospect’s problem rather than their first name.”

That aligns with the repo references and avoids a common failure mode: AI-generated subject lines that look like mass outreach.

Build better personalization

The best usage of the cold-email skill is not “mention something from LinkedIn.” The references argue that personalization should connect to the problem you solve, not just prove you researched the person.

Useful prompt pattern:

• “Use this research signal: they are hiring 3 SDRs. Tie that to likely outbound ramp and follow-up problems, not generic congratulations.”

That produces sharper copy than surface-level personalization such as name-dropping a recent post.

Generate a full sequence, not only the first email

The repo includes concrete follow-up guidance, including cadence and angle rotation. A strong usage request is:

• “Write the initial email plus 4 follow-ups. Each follow-up should add a new angle or value, not just bump the thread. Use day 0, day 3, day 7, day 14, and day 21 timing.”

This matters because a large share of replies come from follow-ups, and the skill is designed with that workflow in mind.

Ask for a self-check before you send

A useful invocation pattern is to ask the agent to evaluate the draft against the skill’s own standards:

• peer-like tone
• concise sentences
• proof used as credibility, not hype
• low-friction CTA
• personalization tied to relevant pain
• no sentence that feels templated or bloated

This is one of the easiest ways to turn the cold-email skill into a repeatable review workflow rather than a one-shot generator.

Practical workflow for teams

A good team workflow looks like this:

1. load product context
2. define segment and outcome
3. provide one proof point and one research signal
4. generate 2–3 variants
5. pick one framework
6. expand to a short follow-up sequence
7. run a quality check
8. adapt by segment, not by individual prospect first

This keeps cold-email usage efficient while preserving message quality.

cold-email skill FAQ

Is this cold-email skill better than a normal prompt?

Usually yes, if your problem is message quality and consistency. The value is not “AI writes emails.” The value is that this skill pushes the agent toward concise structure, problem-linked personalization, proof-based messaging, and realistic follow-up strategy.

Is cold-email for Sales Outreach only?

It is best suited to B2B sales outreach. That includes SDR outbound, founder outbound, agency prospecting, and targeted account outreach. It is less suited to newsletter copy, warm nurture, or customer lifecycle messaging.

Can beginners use this cold-email skill?

Yes, but beginners need to supply more context. If you do not know your audience pain points, proof, or CTA goal, the output will sound competent but generic. The skill helps most when you have at least a basic offer and target defined.

Does the skill help with follow-ups?

Yes. That is one of the stronger reasons to install it. The repository includes dedicated follow-up sequence guidance, angle rotation, and cadence advice instead of treating follow-ups as an afterthought.

Does it cover deliverability or sending tools?

Not really. The cold-email skill focuses on copy, messaging logic, and sequence construction. It does not replace tooling for inbox setup, list building, verification, or sending infrastructure.

When should I not use this skill?

Skip it when:

• you need warm email nurture
• your offer lacks any credible proof
• your audience is too broad to personalize meaningfully
• you want high-volume spammy outreach rather than targeted messaging

The skill is optimized for thoughtful outbound, not brute-force volume.

How to Improve cold-email skill

Give better proof, not more features

The fastest way to improve cold-email output is to provide a real proof point:

• measurable result
• named customer
• time-to-value
• before/after outcome

“Teams improved reply rates by 32% in 6 weeks” is much stronger than “we use AI to optimize outreach.”

Provide research signals with relevance

Better input:

• “They just raised Series B and are hiring outbound reps.”

Worse input:

• “They posted on LinkedIn about leadership.”

The first signal naturally connects to likely sales process strain. The second often produces fake-sounding personalization. The cold-email skill improves when the research signal points to a business problem your offer can address.

Tighten the CTA

A common failure mode is asking for too much too early. The repo’s orientation is toward low-friction asks. Improve outputs by specifying CTAs such as:

• “open to a quick take?”
• “worth sending a short example?”
• “should I share how others handle this?”

These usually fit cold-email usage better than “book a demo.”

Cut length aggressively

The references point toward short emails as a performance lever. If the first draft feels polished but long, ask the agent to:

• cut it below 75 words
• remove throat-clearing
• replace feature claims with one proof point
• keep one idea per sentence

This often improves both realism and reply potential.

Match framework to audience

If results are weak, change the framework before rewriting line by line:

• use QVC for executive brevity
• use PAS when pain is costly and clear
• use BAB when transformation is intuitive

This is a higher-leverage change than swapping adjectives or rewriting the opener repeatedly.

Improve the follow-up angles

If the sequence feels repetitive, assign each email a job:

• initial: tailored observation
• follow-up 1: sharper problem framing
• follow-up 2: proof or benchmark
• follow-up 3: new value asset or angle
• follow-up 4: polite breakup

That mirrors the repository’s follow-up logic and prevents the common “just checking in” trap.

Use the evals to calibrate output quality

Open skills/cold-email/evals/evals.json and compare your prompts against the expected behaviors. The evals reveal what the skill considers good output:

• checking for context files
• sounding peer-to-peer
• choosing a framework
• keeping subject lines short
• giving multiple variants
• using low-friction asks

If your workflow is not producing those traits, fix your prompt before blaming the skill.

Iterate by segment, not only by wording

If performance is poor, do not only rewrite copy. Re-scope the segment:

• narrower company size
• clearer role ownership
• stronger trigger event
• more specific pain

This skill gets better as targeting gets sharper. In cold-email for Sales Outreach, list quality and context quality often matter more than one more copy edit.