agentleFS
cmj-hubcmj-hub/claude-cold-email/skills/cold-email/SKILL.md

cold-email

Skill3 starsChanged 6 months agoMITView on GitHub

Signal-anchored B2B cold email: first touch under 90 words with a binary ask, follow-ups, subject lines, spam lint, domain deliverability, send-list hygiene, reply triage, and outbound audits. Use when the user asks for a cold email, follow-up sequence, deliverability or spam check, reply scoring, or send-list cleanup. Not for give-first offers (sales-offer), picking who to contact (prospect-list), or post-opt-in email (email-sequence).

101 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use free-tools, popups and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • CesarjoquinSkill

    free-tools

    When the user wants to plan, evaluate, or build a free tool for marketing purposes — lead generation, SEO value, or brand awareness. Also use when the user mentions "engineering as marketing," "free tool," "marketing tool," "calculator," "generator," "interactive tool," "lead gen tool," "build a tool for leads," "free resource," "ROI calculator," "grader tool," "audit tool," "should I build a free tool," or "tools for lead gen." Use this whenever someone wants to build something useful and give it away to attract leads or earn links. For downloadable content lead magnets (ebooks, checklists, templates), see lead-magnets.

    Also in 25 projects that use this

  • CesarjoquinSkill

    popups

    When the user wants to create or optimize popups, modals, overlays, slide-ins, or banners for conversion purposes. Also use when the user mentions "exit intent," "popup conversions," "modal optimization," "lead capture popup," "email popup," "announcement banner," "overlay," "collect emails with a popup," "exit popup," "scroll trigger," "sticky bar," or "notification bar." Use this for any overlay or interrupt-style conversion element. For forms outside of popups, see cro. For general page conversion optimization, see cro.

    Also in 23 projects that use this

  • coreyhaines31Skill

    marketing-council

    When the user wants several expert takes on a marketing decision, or asks what a famous marketer would say about it: a simulated board (Seth Godin, David Ogilvy, April Dunford, Byron Sharp, and more), each grounded in their documented frameworks, where they disagree, and a recommendation. To execute, hand off to the relevant skill.

    Also in 18 projects that use this

  • Hussain-ElfalahSkill

    andreessen

    Also in 4 projects that use this

  • Hussain-ElfalahSkill

    business-investment-advisor

    Also in 4 projects that use this

  • roerohanSkill

    emails

    When the user wants to create or optimize an email sequence, drip campaign, automated email flow, or lifecycle email program. Also use when the user mentions "email sequence," "drip campaign," "nurture sequence," "onboarding emails," "welcome sequence," "re-engagement emails," "email automation," "lifecycle emails," "trigger-based emails," "email funnel," "email workflow," "what emails should I send," "welcome series," or "email cadence." Use this for any multi-email automated flow. For cold outreach emails, see cold-email. For in-app onboarding, see onboarding.

    Also in 24 projects that use this

  • +5 more from Cesarjoquin
  • +1 more from Hussain-Elfalah

The file

Show its rules and text

We split this skill into the rules it gives an agent

15 rules. 0 are common, 6 are distinctive, and 9 are very specific to this project.

  • 0

    Common

    Most of the 94 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 9

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Run this before routing anywhere.

    line 21
  2. Very specific:

    Look for brand-config.json and SOUL.md in the operator's project root.

    line 24
  3. Very specific:

    If either is missing and the operator did not pass --no-config, run the setup mode (modes/setup.md) before drafting real outreach.

    line 26
  4. Very specific:

    If brand-config.json exists, sanity-check the fields the requested job needs (e.g.

    line 31
  5. Very specific:

    If the psp or evp block is missing, say which pack produces it (/plugin install psp@gtm-operator-skills, /plugin install …

    line 34
  6. Distinctive:

    Treat SOUL.md as voice only.

    line 37
  7. Distinctive:

    Run the preflight above first, every time.

    line 42
  8. Very specific:

    If $ARGUMENTS starts with a mode name below, go straight to that mode (the rest of $ARGUMENTS is its input).

    line 44
  9. Very specific:

    If $ARGUMENTS is empty, run status: it reads the project and names the one next step.

    line 46
  10. Distinctive:

    Read the mode file with the Read tool and follow it.

    line 51
  11. Distinctive:

    A cold email is a hire — not a search query: Every prompt anchors on a Pain Signal Profile, names the constraint, and forces the output …

    line 78
  12. Distinctive:

    If any are missing, ask the user — don't fabricate.

    line 109
  13. Very specific:

    Write the draft to gtm/letter.json, then:

    line 142
  14. Very specific:

    Every scorer takes --file PATH or --stdin, prints text by default and one JSON object with --json, and exits 0 pass, 1 refused (each reason …

    line 164
  15. Very specific:

    If a companion pack is not installed, name it and its install line (/plugin install <name>@gtm-operator-skills).

    line 188

“Common” is measured: kept by at least 6 in 10 of the 94 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Cold Email — JMC Outreach Craft Skill

Comprehensive cold email orchestrator for B2B operators. Anchors every output on a Pain Signal Profile (not demographics) and forces the output into a shape you can ship.

Preflight (every invocation)

Scoring something pasted needs no setup. If the operator handed you a line, post, draft, or file to score, run the scorer on it first and report the result; missing config only means some checks are skipped, so say which. Offer setup afterwards as the next step. Check whether files exist with Read or Glob, not a shell command.

Run this before routing anywhere. It is what AGENTS.md rules 1 and 2 require.

  1. Look for brand-config.json and SOUL.md in the operator's project root. Read both if present.
  2. If either is missing and the operator did not pass --no-config, run the setup mode (modes/setup.md) before drafting real outreach. If operator or icp is missing, say "Run /gtm:setup once for the whole suite" first. Offer once; if they decline, say the output will be framework-shaped but generic.
  3. If brand-config.json exists, sanity-check the fields the requested job needs (e.g. evp.primary for drafting, infrastructure.sending_domain for deliverability). Ask for anything missing. Never fill it in. If the psp or evp block is missing, say which pack produces it (/plugin install psp@gtm-operator-skills, /plugin install evp@gtm-operator-skills) instead of inventing a signal, pain, or EVP.
  4. Treat SOUL.md as voice only. It can change word choice; it cannot lift the word limit, soften the binary CTA, or un-ban a pattern.

Modes — route by $ARGUMENTS

Run the preflight above first, every time. Then pick one mode:

  • If $ARGUMENTS starts with a mode name below, go straight to that mode (the rest of $ARGUMENTS is its input).
  • If $ARGUMENTS is empty, run status: it reads the project and names the one next step.
  • Otherwise match the user's words to the table. If intent is still ambiguous, ask one clarifying question — never two.

Read the mode file with the Read tool and follow it.

You say / argumentMode file
(nothing), status, "where do I start", "what's next"modes/status.md
setup, "set up my brand config", "configure my voice"modes/setup.md
write, "write a cold email to...", "rewrite this cold email"modes/craft.md
sequence, "build a follow-up sequence", "Day 3 / 7 / 14"modes/craft.md (sequence mode)
subject, "subject lines for...", "critique this subject"modes/subject.md
lint, "spam-check this", "will this land in spam"modes/lint.md
deliverability, "is my domain ready", SPF / DKIM / DMARCmodes/deliverability.md
list, "score / clean my send list", "dedup this CSV"modes/list.md
reply, "score this reply", "triage my replies"modes/reply.md
nurture, "re-engage cold or stalled prospects" (not opted in)modes/nurture.md
audit, "audit my outbound program"modes/audit.md
rhythm, "this week's queue", "Friday review"modes/rhythm.md
"Review / score this draft"cold-email-reviewer agent

Files the modes write in the operator's project go under gtm/ at the project root (create it if missing): gtm/letter.json (the draft: public_signal, subject, letter), gtm/send-list.csv (the send list; --write adds .cleaned.csv and .removed.csv beside it), and gtm/replies.jsonl (one reply per line). brand-config.json and SOUL.md stay at the project root.

Core principles (the JMC stance)

  1. A cold email is a hire — not a search query. Every prompt anchors on a Pain Signal Profile, names the constraint, and forces the output into a shape you can ship.
  2. No demographics in the opener. "VP of Marketing at Series B SaaS" isn't a signal. "Just posted a job for a Demand Gen Lead" is.
  3. Binary CTAs only. "Worth a look?" or "Is the 12-day close on your list this quarter?" not "Let me know if you'd like to learn more." Name a day or time only when the operator gave you their availability; never invent a slot.
  4. <90 words for the first touch. Anything longer signals desperation.
  5. One specific ask per email. Stack-ranked asks confuse buyers.
  6. No "Hope you're well" / "Just bumping this". Banned openers.
  7. Found is not sendable until verify = deliverable. A scraped or enriched address stays off the send list until verification returns deliverable. Catch-all, unknown, and invalid are not "valid."

Variables (every mode uses these)

The core framework operates on seven variables. Capture them once, reuse across every mode:

VariableSourceExample
firstNameRecipient listSarah
roleLinkedIn / CRMVP of Demand Gen
companySameAcme Cloud
signalPublic signal — job post, funding, product launch, leadership change"Just posted 'Senior Demand Gen Lead' role on LinkedIn 4 days ago"
painPain Signal Profile — what the signal implies"Pipeline gap; existing demand gen motion isn't producing enough SQLs"
evpYour one-line value prop in their words"We help Series-B SaaS rebuild first-touch email around public signals without changing their sequencer"
senderNameYou (brand-config.operator.name)Jay

If any are missing, ask the user — don't fabricate.

Self-check rubric (run on every output before showing it)

AGENTS.md rule 5 points here. A draft that fails any hard check is regenerated, not shown with a caveat.

Hard checks — first touch

  • Body under 90 words (T2/T3 under 70, T4 under 60)
  • Line 1 names the signal in the source's own words, and the signal is something they did, not who they are
  • Quotation marks only around words the source itself quotes or contains; a paraphrase is never dressed as a quote
  • No day, time, asset or offer the operator did not give you (no "Thursday", no "the attached sample", no "a short write-up")
  • No congratulation, compliment or "saw that you" opener, in any touch
  • Plain salutation and sign-off (Dana, and [Your name] when the sender is not given); no em dashes; no notes to the operator unless something was refused
  • Pain is one sentence in the buyer's vocabulary (brand-config.psp.vocabulary when present)
  • EVP is 22 words or fewer, names one outcome and one tradeoff
  • Exactly one ask, answerable yes / no
  • No pattern from references/banned-patterns.md or brand-config.tone.banned_phrases / SOUL.md "never" list
  • Subject 7 words or fewer; no ALL CAPS, emoji, clickbait, or fake Re: / Fwd:
  • No unmerged tokens ({{firstName}}, <company>)
  • Nothing invented: every name, signal, metric, and case study came from the operator or a source they gave you

Deterministic checks — run when Bash is available

Write the draft to gtm/letter.json, then:

  • score_letter.py --file gtm/letter.json: exit 0 (signal quoted, no demographics, under 90 words, one yes/no ask)
  • spam_word_lint.py --file gtm/letter.json: score 75 or higher
  • score_subject_line.py --file gtm/letter.json: score 70 or higher

On exit 1 each reason reads - what is wrong → what to change. Apply every change and run it again; do not show a draft that fails.

Flag, don't hide — if the signal, pain, EVP, or vocabulary was a guess, say which line is uncertain and what data would settle it (AGENTS.md rule 10).

Running the bundled scripts

The scripts live in scripts/ at the plugin root. Call them as ${CLAUDE_PLUGIN_ROOT}/scripts/<name>.py, which resolves to the installed plugin from any project directory. Outside Claude Code, python3 scripts/<name>.py from the repo root is the same script. All are Python 3.8+ stdlib only; DNS checks need dig.

Every scorer takes --file PATH or --stdin, prints text by default and one JSON object with --json, and exits 0 pass, 1 refused (each reason with its fix, then Next: fix the lines above and run this again.), 2 bad input. On a pass the last line names the next step.

ScriptMode
score_letter.pycraft, self-check (before the spam lint)
spam_word_lint.pylint, self-check
score_subject_line.pysubject, self-check
score_reply.pyreply
score_list.pylist
check_deliverability.py, dig_dns.shdeliverability

Works with the suite

This is step 4 of the GTM operator suite (/plugin marketplace add cmj-hub/gtm-operator-skills).

  • Reads: psp, evp, icp, tone, infrastructure, operations from brand-config.json if present.
  • Writes: tone, infrastructure, operations. Merge at the field level; never overwrite another pack's keys.
  • Before this: psp (/psp:psp) and evp (/evp:evp), when there is no psp or evp block; prospect-list (/prospect-list:who-to-contact), when you do not yet know who to contact.
  • Instead of this: sales-offer (/sales-offer:cold-offer), when the first touch should hand over a leak and a prototype rather than ask.
  • After this: email-sequence (/email-sequence:lifecycle-email), when a prospect replies and opts in. End a successful run with that Next: line.

If a companion pack is not installed, name it and its install line (/plugin install <name>@gtm-operator-skills); do not do its job inline.

References

Load these on demand for deeper context:

  • references/jmc-framework.md — The full JMC cold email framework
  • references/banned-patterns.md — Lines / openers / closes that fail
  • references/binary-ctas.md — 30 binary CTA patterns
  • references/psp-anchors.md — How to derive signal → pain → EVP
  • references/bulk-sender-rules.md — Google / Yahoo / Microsoft sender requirements
  • references/audit-rubric.md — The 30-point outbound audit rubric
  • SOUL.md — Operator voice template (the shape the setup mode fills in)
  • examples/t1.email.md — A first touch that passes every check (lints at 100)
  • examples/spam.email.md — A draft the spam lint rejects

Specialist agents

  • cold-email-reviewer — Quality scorer (0-100) against framework
  • cold-email-deliverability-auditor — DNS/reputation specialist

Free hosted version

The same job runs in a browser, no install and no key: Cold Email Linter

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.