Outreach Architect

What this skill does

• **ICP first, and narrow:** firmographics + technographics + pain + a timing trigger; reverse-engineered from your 10–20 best customers (a too-broad ICP is the #1 cause of silence)
• **A real trigger before the first sentence:** funding, a new exec, a relevant job post, a tech change — reach out within 24–48h; signal-based emails reply 3–5× better
• **Human copy:** PAS, ≤90 words, situational opener, one soft CTA, no AI slop (tells stripped)
• **Multi-touch cadence:** 4–7 touches across email/LinkedIn/phone, each follow-up with a distinct job (value-add, proof, new angle, breakup)
• **Relevant, not creepy personalization:** the 30-second-LinkedIn test + first-meeting test, one real signal, the 70/30 rule
• **Deliverability as a gate:** dedicated warmed domain, SPF/DKIM/DMARC, volume caps, plain text, a spam linter (with AI-slop detection)
• **Region-aware compliance:** CAN-SPAM, GDPR, UWG §7, CASL, Spam Act — consent/opt-out, honest about the hard cases
• **Two Node scripts:** spam/deliverability linter and a cadence planner (dated multi-touch sequence)

Description

Outreach Architect is a skill for the whole cold-outreach journey — it turns an **ICP plus your offer** into a personalized, non-spammy **B2B outreach sequence**. The difference: it's the **strategy, copy and compliance brain** the hosted sending tools (Instantly, Apollo, Lemlist, Outreach) assume you'll bring but don't provide — and it deliberately **doesn't send** (no autonomous AI spam; a human stays in the loop). Founding rule: **relevance beats volume, and beats craft.** A single true observation about *this* company outperforms a perfectly-written generic template 3–5× — in 2026 cold outreach is resonance, not reach; reply rates fell to ~3% because AI made generic blasting free. So it starts with **strategy**: a tight ICP (firmographics + technographics + pain + a timing **trigger** like funding, a new exec, a relevant job posting) and the buyer persona, then the **copy** — PAS for the first touch, ≤90 words, a **situational opener** ("Saw [fact] and figured [implication]") not a compliment, **don't pitch** (pitching cuts replies up to 57%), one light proof, **one soft CTA** — with every AI tell stripped. Then the **cadence**: 4–7 touches across email/LinkedIn/phone where each follow-up does a new job (value-add → proof → new angle → breakup), because 42–70% of replies come from touches 2–4. And two hard gates: **deliverability** (dedicated warmed domain, SPF/DKIM/DMARC, ~25–40 sends/inbox/day, plain text, a spam-word check) and **region-aware compliance** (US CAN-SPAM needs no consent — Germany via UWG §7, Canada via CASL, Australia do; classic cold email is largely unlawful in Germany). You walk away with **durable artifacts** — an ICP doc, the full sequence, a personalization playbook, subject/opener A/B variants and a compliance checklist — that you paste into your own sender, edit into your voice, and approve. Honest: AI does the research and structure; you own voice, truth and the send.

Examples

Tested with Claude Code

subject: congrats on the series b

Hi {{First}},

Saw {{Company}} raised its Series B and is bringing on five SDRs — that ramp usually hits a wall fast, new reps taking 4–5 months to quota.

We cut that to ~10 weeks for {{one comparable SaaS}} by automating their first-90-days playbook.

Worth a quick look at how?

{{You}}

What it does not do

• **Doesn't send** and doesn't scrape/buy/verify leads or warm up a domain — brain, not sender; you run the output through your own tool
• **Not legal advice** — region-aware general info; hard cases Germany (UWG §7)/Canada (CASL) to counsel or the official source
• **No autonomous mode** — human-in-the-loop; you edit into your voice, verify facts, approve
• **No guaranteed replies or inbox placement** — it raises the odds, it can't promise outcomes

Compatibility & tech

Tested (internal)

3 scenarios

Recommended runtime

Claude Opus or Sonnet; with web access to verify triggers and current sender/legal rules (Google/Microsoft, CAN-SPAM/GDPR/UWG/CASL) live. The two scripts need Node (no extra packages); without Node the skill produces the spam lint and cadence by hand from the references.

Modes

Define the ICP / target audience · Write a cold email (+ A/B variants) · Build the multi-touch sequence / cadence · Personalization research · Deliverability & compliance check

Inputs

offer / value prop + target country · a rough ICP or your 10–20 best existing customers · a trigger / occasion (funding, hiring, a tech change) · an existing email draft or sequence (to improve)

Output format

A Markdown deliverable per mode, copy-paste ready — an ICP doc, the full sequence (copy + dated cadence), a single email plus variants, or a checklist. Copy-paste content (an email, a cadence, the ICP doc) sits in its own clean block.

Subcategory

Cold outreach, prospecting & customer acquisition

License

Proprietär

Security profile

What you get

• akquise-architekt-1.0.0/10 files
  • .claude-plugin/marketplace.json
  • plugins/akquise-architekt/skills/akquise-architekt/10 files

    • SKILL.md
    • manifest.json
    • references/6 files

      • akquise-architekt-compliance.md
      • akquise-architekt-copy.md
      • akquise-architekt-deliverability.md
      • akquise-architekt-icp.md
      • akquise-architekt-personalization.md
      • akquise-architekt-sequence.md
    • scripts/2 files

      • cadence-planner.mjs
      • email-spam-check.mjs
  • .agents/skills/akquise-architekt/→ universal — same content (Codex, Cursor, Copilot, Gemini, Windsurf, Cline)
  • LICENSE.txt

Installation

Also works as a chat prompt

Note

Outreach Architect is a first, practical guide and **not legal advice**: it gives region-aware general information on anti-spam and privacy law (CAN-SPAM, GDPR, **UWG §7**, CASL, the Spam Act) that changes and varies by country — verify the official source live and get counsel for the hard cases (Germany, Canada). **In Germany, classic cold email generally requires prior consent under UWG §7** and is largely unlawful; the skill switches to consent-safe routes there. It doesn't send anything, scrape or buy leads, or guarantee replies or inbox placement — you stay the human who reviews and sends. The bundled scripts need Node (no extra packages).

Changelog

• v1.0.021.08.2026Initial release: turns an ICP + offer into a personalized, non-spammy B2B outreach sequence. Tight targeting + a real trigger before the first sentence, human-reading copy (PAS/AIDA, ≤90 words, situational opener, one soft CTA, AI tells stripped), a 4–7 touch cadence across email/LinkedIn/phone (each follow-up with a new job), plus a deliverability gate (SPF/DKIM/DMARC, warmed domain, volume caps, spam check) and region-aware compliance (CAN-SPAM, GDPR, Germany's UWG §7, CASL). Durable artifacts (ICP doc, sequence, personalization playbook, A/B variants, compliance checklist) to paste into your own sending tool. Two Node scripts (spam linter with AI-slop detection, cadence planner), 6 references. It doesn't send — the strategy/copy/compliance brain, human-in-the-loop.

Frequently asked questions