A 6-skill cold outreach engine for Claude Code. It merges three layers into every email: your business strategy (the trust-ladder doctrine), GEX prospect research signals, and short-form anti-AI voice. The whole sequence is built to earn one positive reply, not an open. Each touch sells the next small, reversible step instead of asking a stranger for the leap, and nothing ships until it clears a QA gate.
Install from the Mindantic marketplace:
/plugin marketplace add mindantic/claude-plugins
/plugin install outreach@mindantic
Then restart Claude Code so the skills load. After that, the /outreach-* commands are available in any session.
Local development (running this repo directly from a clone):
claude plugin marketplace add /path/to/outreach
claude plugin install outreach@mindantic-local
New campaign? Start with /outreach-planning. It builds the four context files everything else reads, then each skill hands off to the next in the pipeline.
/outreach-planning # once per offer: build the strategic foundation
/outreach-research # optional: research a named prospect (raises reply rates)
/outreach-writing # draft the 4-touch sequence with A/B tracks
/outreach-lead-magnet # build the asset the emails promise (before launch)
/outreach-reviewing # QA gate: no sequence ships until this passes
Already have opted-in recipients (signups, trials, subscribers)? That is a different game. Use /outreach-lifecycle instead.
outreach-planning -> outreach-research -> outreach-writing -> outreach-lead-magnet -> outreach-reviewing
(foundation) (optional) (draft) (build asset) (QA gate)
separate track: outreach-lifecycle (warm, opted-in recipients only)
- outreach-planning: Set up a campaign by diagnosing the offer, mapping provable proof, building ICP and objection profiles, and producing the four context files all downstream skills consume. Runs once per offer.
- outreach-research: Run the GEX 10-minute tiered signal hunt on one prospect, ranking operational truths, business signals, and social proof by the personalization hierarchy. Optional but recommended.
- outreach-writing: Write a full 4-touch cold sequence (Day 1 / 3 / 7 / 12) with both A/B CTA tracks, enforcing the short-form anti-AI copy rules and every doctrine constraint.
- outreach-lead-magnet: Build the real deliverable the emails promise and confirm it matches the promise exactly. A real thing the prospect keeps, not a teaser, and it must exist before launch.
- outreach-reviewing: QA a sequence against the full merged checklist (spam words, banned phrases, positive-reply test, format, integrity, deliverability) and lock a send date. The mandatory gate between writing and sending.
- outreach-lifecycle: Write nurture, onboarding, or newsletter sequences for opted-in recipients. Longer and metric-rich, with its own rules separate from cold outreach.
A typical campaign runs the pipeline once, then writing/reviewing per batch:
- Run
/outreach-planningand answer the questions. It writescontexts/OFFER.md,ICP.md,PROOF.md, andDELIVERABILITY.md. Do this once per offer. - For each prospect worth the effort, run
/outreach-research. It produces a ranked-signal file undercampaigns/<campaign>/RESEARCH/. Skip a prospect if no real signal turns up. - Run
/outreach-writing. It drafts the full 4-touch sequence in both A/B tracks tocampaigns/<campaign>/SEQUENCE.md. - Run
/outreach-lead-magnetto build the asset the emails promise, before you launch. - Run
/outreach-reviewing. It runs the QA gate, applies fixes, produces a deliverability checklist, and locks a send date. Nothing ships until this passes.
For warm, opted-in recipients, ignore the pipeline above and run /outreach-lifecycle instead.
A few headline rules. reference/DOCTRINE.md is the source of truth and wins on any conflict.
- One job: earn one reply. Sell the next rung of the trust ladder, never the leap.
- Short-form copy. 3 to 4 sentences, 40 to 80 words, ending in one positive-reply question. No em-dashes, no cliches.
- Link discipline. No link in Touch 1. One bare-domain link in Touch 2 only.
- The 4-touch spine over about 12 days. Day 1, Day 3 in-thread, Day 7 and Day 12 with new subjects. Each touch has a different job.
- Deliverability decides opens, not copy. Configure SPF, DKIM, and DMARC, send from a separate domain warmed for 4 to 6 weeks, cap at 25 to 30 sends per day per mailbox, plain text only.
- Targets. Opens 40 to 60 percent, replies 8 to 15 percent, with a third to a half of replies positive.