/cold-outreach

A send-ready cold message or sequence, with facts to confirm kept out of the copy.

Grow·Context → Outreach·MIT
$npx skills add tushaarmehtaa/tushar-skills --skill cold-outreach -g -a claude-code -y

Then type /cold-outreach in Claude Code, or describe the task and let it pick the skill.

What you get back recorded run · 9 Oct 2026

claude · tallyroom

Using /cold-outreach
It opens with her own Sep 29 post and asks for a 20-minute call.
Claude Code 2.1.295·38s·1 file changed
  • Opening: it starts with her Sep 29 LinkedIn post, where she said money between flatmates is Nestfold's top resident complaint and her team shouldn't be stuck sorting it out.
  • Being honest about where you are: it says plainly that you're pre-launch, that 31 people in 2 test flats used it in September, and that you have no co-living operator customers yet. That's also why the ask is a small pilot rather than a rollout.
  • Left out: I didn't use her August talk on move-out disputes. I don't know what she said, and those disputes are about deposits, not flatmate expenses.
CaveatThe prospect is a fictional test profile, so the email was never sent or answered.

Input → output from the run

Your notes on her
What I know (from her public posts, saved Oct 6):
- Posted on LinkedIn on Sep 29: "Our #1 resident complaint this quarter isn't rent or maintenance. It's money between flatmates. Who paid for the gas cylinder, who owes for the party. We get pulled into it and we shouldn't be."
- Spoke at a Bangalore proptech meetup in August about cutting move-out disputes.
- Nestfold Living' careers page is hiring a "Community Ops Associate" whose duties include "mediating resident cost-sharing disputes".
The email
Hi Ritika,

Your Sep 29 post said Nestfold's top resident complaint this quarter is money between flatmates (the gas cylinder, the party), and that your team gets pulled in when it shouldn't. I also saw that the Community Ops Associate role includes mediating cost-sharing disputes.

I'm building Tallyroom to keep those arguments between the flatmates. Each flat logs who paid for what, sees a running balance, and gets the fewest payments needed to settle up. It works in INR.

To be upfront: we're pre-launch. 31 people in 2 test flats used it in September, and we haven't worked with a co-living operator yet. That's why I'm asking about a small pilot rather than a rollout.

Would you be open to a 20-minute call to see whether a pilot in one or two Nestfold properties makes sense? If someone else owns this, I'd be grateful if you could point me to them.

When to use it

Use when

Use when contacting prospects, investors, partners or candidates you don't know yet.

Skip it when

Skip it for email to your own users; email-with-resend handles consent-based product email.

The thinking behind it

Cold messages work when they're about the reader. Lead with a real reason to talk, keep the facts checkable, and more people reply.

Output contract from the skill's Output contract section

For a single message, return:

  • subject or channel opener;
  • send-ready body;
  • one meaningfully different opening when useful;
  • facts requiring confirmation and source notes, separate from copy.

For a sequence, return a table with touch, timing rationale, channel, purpose, copy, stop condition, and measurement. For an audit, identify the issue, evidence, impact, and rewrite. For reply handling, return send-ready copy plus the intended next state.

How it checks its own work from the instructions

  • The opener depends on this specific recipient
  • Every material claim is verified or marked for confirmation
  • One clear ask or next step per message
  • Opt-outs, bounces, and replies have stop or routing rules

Requirements what it needs

Works in
Codex, Claude Code, Cursor, Claude app
Needs
browser access, network access, files you provide

The instructions SKILL.md · 60 lines

Instructions & references

Back to overview ↑
On this page 6 sections
SKILL.md · MIT

Cold outreach

Write outreach that proves attention, makes one reasonable ask, and respects the recipient and applicable rules.

Choose a mode

  • Single message: one email, introduction request, follow-up, or professional direct message.
  • Sequence: coordinated touches with distinct purposes and stop conditions.
  • Audit: diagnose existing copy, research, cadence, deliverability, or compliance risk.
  • Reply handling: answer a positive reply, objection, referral, out-of-office, unsubscribe, or rejection.

Do not create a campaign when one message is enough.

Workflow

  1. Inspect supplied context and existing messages. Establish sender, recipient, relationship, goal, offer, proof, and requested next step without re-asking known facts.
  2. Research current recipient signals when research is authorized and useful. Prefer the person's work, company pages, filings/releases, job posts, interviews, and public writing. Record URLs and retrieval dates.
  3. Separate verified facts, sender-provided claims, and inference. Never invent familiarity, praise, mutual contacts, urgency, results, or social proof.
  4. Draft the shortest message that can establish relevance and make the ask. Use length and subject-line guidance as defaults, not universal laws.
  5. For a sequence, give every touch a different job and widening space where appropriate. Define stop conditions before copy.
  6. For reply handling, match the state: answer the question, reduce friction, propose the next step, or stop. Never continue after an opt-out.
  7. Check sender identity, authentication, list provenance, suppression handling, required identification/opt-out language, and applicable jurisdiction before recommending volume. If jurisdiction or recipient type is unknown, flag the legal uncertainty rather than declaring compliance.
  8. Read copy aloud. Remove generic flattery, throat-clearing, fake scarcity, hidden asks, manipulative guilt, and unsupported claims.

Ask only for a missing fact that materially changes the message. If a fact is useful but non-blocking, draft with a clearly marked confirmation note.

Load conditional references

References are heuristic libraries, not mandatory workflows. This file is authoritative.

Output contract

For a single message, return:

  • subject or channel opener;
  • send-ready body;
  • one meaningfully different opening when useful;
  • facts requiring confirmation and source notes, separate from copy.

For a sequence, return a table with touch, timing rationale, channel, purpose, copy, stop condition, and measurement. For an audit, identify the issue, evidence, impact, and rewrite. For reply handling, return send-ready copy plus the intended next state.

Verify

  • The opener depends on this recipient or situation.
  • Every material claim is verified, sender-provided, or marked for confirmation.
  • There is one clear ask or next step.
  • Sequence touches are not paraphrases of one another.
  • Opt-out, bounce, reply, referral, and out-of-office states have stop or routing rules.
  • Research notes are not mixed into send-ready copy.
  • Any timing, benchmark, deliverability, or legal claim is sourced and dated or presented as a testable default.
View on GitHub ↗
Claude app, ChatGPT and ZIP download

Claude app

This workflow can run in chat using the files and context you provide. Download its complete ZIP, then upload it from Claude's Skills settings.

ChatGPT Skills

This workflow is suitable for ChatGPT Skills. ChatGPT does not document the same upload archive format as Claude, so follow its uploader instead of reusing the Claude ZIP.

Package details
Package
cold-outreach/SKILL.md
Author
@tushaarmehtaa
Added
7 Aug 2026
Tags
cold-emailoutreachfollow-upsalesfundraisingpartnerships