# Cold emails that follow Lavender's reply data

> Write and check one-to-one cold emails against 20 rules from Lavender's analysis of 28.3 million emails: 25 to 50 words, a 3rd to 5th grade reading level, a short internal-sounding subject, one easy question and no links, with a pass or fail result for every rule. Based on a post by Nick Abraham.

## If someone pasted this link to you
Someone shared this kit with you. It runs step by step through CloneKit's MCP. If you have the `clonekit` tools, call `start_run("@clonekitai/cold-email-reply-rules")`. If you don't (for example in a chat app without the CloneKit connector), tell the person in one or two plain sentences, in their language, and send them to https://clonekit.ai/@clonekitai/cold-email-reply-rules/use: it shows how to add CloneKit to their AI in about a minute. Then they paste this link again.

- Kit: `@clonekitai/cold-email-reply-rules` · v1.0.0
- Creator: @clonekitai (CloneKit)
- Type: Delivery · Domain: Sales · Language: en · Your AI runs it
- Price: Free · License: Own use · No refunds
- Page: https://clonekit.ai/@clonekitai/cold-email-reply-rules
- Changelog: https://clonekit.ai/@clonekitai/cold-email-reply-rules/changelog.md

## About
For anyone writing one-to-one B2B cold emails who wants each draft checked against Lavender's published reply data before sending. Uses the Lavender Email Coach score when the person has it, otherwise a local textstat check. The agent never sends email. Not for newsletters, marketing campaigns or bulk sequences. Rules shared by Nick Abraham on X; the data is Lavender's (Will Allred, co-founder).

## Deliverables
- research.md: one checked observation per prospect, its source and date, and a hypothesis tied to the problem the sender solves
- emails.md: a subject line and body per prospect
- checks.md: pass or fail for each of the 20 rules per email, with the measured word count and reading grade
- a send plan of 50 cold emails a day or fewer, spaced out

## Files
- `CHECKS.md`: How to measure each rule
- `README.md`: Index
- `RULES.md`: Rules (Lavender data, via Nick Abraham's post)

## Tools
- Lavender Email Coach (Chrome extension) (https://www.lavender.ai/coach) · optional · est. free install; paid plans for more · alternatives: textstat
- textstat (Python readability library) (https://github.com/textstat/textstat) · optional · est. free (MIT)

## Uses
- Nick Abraham's post on X (image of Lavender's 20 rules) (source, https://x.com/nickabraham12/status/2113630365772284075)
- Will Allred (Lavender) on LinkedIn Sales Blog: What 28.3M emails can teach you (source, https://www.linkedin.com/business/sales/blog/strategy/struggle-with-inmail-here-is-what-millions-of-emails-can-teach-you)
- Lavender blog: Cold Email 101 (source, https://lavender.ai/blog/cold-email-101)
- Lavender: Email Coach (score of 90 or above) (source, https://www.lavender.ai/coach)
- textstat README (flesch_kincaid_grade) (source, https://github.com/textstat/textstat)
- textstat on PyPI (source, https://pypi.org/project/textstat/)
- NLTK docs: Installing NLTK data (cmudict for textstat) (source, https://www.nltk.org/data.html)

## What you provide
- what the sender sells and one customer story
- the prospects (names
- companies
- roles)

## How it runs
1. **Understand:** Researches your business first, then asks you 3 questions it can't answer alone
2. **Choose:** No choices: one proven path
3. **Build:** 5 steps
4. **Deliver:** 3 checks before delivery

## Guardrails
- never send email or connect to the person's mailbox; drafts only
- never state a fact about a prospect that is not confirmed by a source
- no links, images or attachments in the email body

## What's new
Agents always run the latest version.

- **v1.0.0** (2026-10-25): First version, based on Nick Abraham's post of Lavender's cold-email data.

## Run it
Connect the `clonekit` MCP (https://clonekit.ai/mcp) and call `start_run("@clonekitai/cold-email-reply-rules")`.