AI isn't replacing the SDR desk, it's finishing it. The gtm-skills SDR pack covers the whole desk end-to-end, and the reply-handling additions close the one gap that mattered: nothing in the catalog handled inbound replies until now. Build the list, write the message, book the call - and now sort the replies, clear the inbox, and save the no-shows. Ten skills, one job description, free and open source.
"Will AI replace sales jobs" gets searched a lot, and the honest answer for SDRs is no, not the job. What's actually happening is narrower and more useful: the mechanical parts of the desk - build the list, write the opener, sort the replies - are getting handed to AI, one job at a time, while the decisions that matter stay with a person.
gtm-skills is a free, MIT-licensed collection of Claude Skills organized around real GTM job roles. The SDR pack covers the whole desk end-to-end. The reply-handling additions close a real gap: across the entire catalog, nothing handled inbound replies until they landed. You could build a list, write a cold email, and book a call, but the moment a prospect wrote back, you were on your own. Not anymore.
TL;DR
- The gtm-skills SDR pack covers the whole desk end-to-end, and reply-handling skills close the gap that used to exist across the catalog.
- Reply-handling skills: the-reply-classifier, the-inbox-zero-enforcer, the-list-cleaner, the-call-booker, the-no-show-save.
- The rest cover list building, ICP scoring, cold email, subject lines, and sequence auditing.
- Install once with npx skills add sidchaudhary/gtm-skills, then paste the one you need into a Claude project.
- The send button, the decision to contact someone, and any angry reply still go through a human, every time.
How to install these in about 60 seconds
- Run npx skills add sidchaudhary/gtm-skills, or clone the repo directly.
- Open Claude and create a project. Name it after the skill, for example "Reply Classifier."
- Paste the skill's SKILL.md into the project instructions.
- Start a chat inside that project and give it your inputs.
Works in Claude chat, in a Claude Project, or in Claude Code. Nothing else to install.
The 10 skills
Ten skills covering every part of the SDR desk: three for building, cleaning, and scoring the list, four for writing the message and booking the meeting, and three reply-handling skills that finally close the inbound gap the catalog never covered before.
| Skill | What it does | Status |
|---|---|---|
| 1. The List Builder | Builds and qualifies a prospect list from an ICP definition, scored Hot/Warm/Cold/Skip | Existing |
| 2. The Fit Scorer | Scores an account list against your ICP into High/Medium/Low priority tiers | Existing |
| 3. The List Cleaner | Dedupes and flags a raw list before it hits a sequence: wrong titles, stale roles, broken data | New this session |
| 4. The Cold Opener | Writes a complete cold email body under 120 words from research and a trigger signal | Existing |
| 5. The Subject Line Lab | Generates five scored subject line variants with a recommendation | Existing |
| 6. The Sequence Doctor | Audits a cold sequence and rewrites the three weakest emails | Existing |
| 7. The Call Booker | Writes the booking message, reschedule, confirmation, and day-before reminder | New this session |
| 8. The Reply Classifier | Sorts an inbound reply into Interested/Later/Referred/Objection/Dead/Angry with evidence | New this session |
| 9. The Inbox Zero Enforcer | Sorts a full inbox into Drafted/Escalated/Scheduled/Closed | New this session |
| 10. The No-Show Save | Runs a 3-stage recovery sequence for a missed meeting | New this session |
Build, clean, and score the list
Three skills cover the input side of the desk: getting a list that's worth writing to in the first place.
1. The List Builder
Builds and qualifies a prospect list from an ICP definition, sourced from data you provide or public web research, scored Hot/Warm/Cold/Skip with evidence and confidence per lead. Feeds directly into the-fit-scorer and the-list-cleaner. Read the skill.
2. The Fit Scorer
Scores an account list against your ICP criteria and returns a priority tier - High, Medium, or Low - with a one-sentence rationale per account. Use it before an outbound campaign so the good hours go to the good accounts, not the whole list. Read the skill.
3. The List Cleaner
New this session. Dedupes and flags a raw list before it ever hits a sequence: wrong titles, stale roles, wrong companies, broken emails and phone numbers. This is the step between the-fit-scorer and the-cold-opener that most desks skip, and the reason a scored list still bounces half its sends. Read the skill.
Write and book
Once the list is clean and scored, four skills carry a name through the message, the send-ready check, and getting the meeting on the calendar.
4. The Cold Opener
Takes prospect research, a specific trigger signal, and a product value prop, and writes a complete cold email body under 120 words, personalized to the signal. Pairs with the-subject-line-lab for the send. Read the skill.
5. The Subject Line Lab
Takes a cold email body and target persona and generates five scored subject line variants with a recommendation. Run this before sending any sequence at scale, not after the open rates come back disappointing. Read the skill.
6. The Sequence Doctor
Audits a cold outbound sequence and returns a gap analysis per email, rewrites for the three weakest emails, and a recommendation for any missing step. This is the one you reach for when reply rates drop and you need to know why, not just that they did. Read the skill.
7. The Call Booker
New this session. Writes the booking message, the reschedule, the confirmation, and the day-before reminder, so the meeting actually shows up on the calendar instead of dying in a "sounds good, let's find time" reply. Read the skill.
Classify and clear replies
The three skills that close the gap. Before this session, once a prospect wrote back, there was no skill in the catalog built to handle it.
8. The Reply Classifier
New this session. Sorts an inbound reply into Interested, Later, Referred, Objection, Dead, or Angry, with the evidence for that call and a recommended next action. This is the skill that used to not exist anywhere in the catalog. Read the skill.
9. The Inbox Zero Enforcer
New this session. Processes a full inbox in one pass and sorts every thread into Drafted, Escalated, Scheduled, or Closed. Built for the morning after a sequence goes out, when forty replies land overnight and half of them need a different response than the other half. Read the skill.
10. The No-Show Save
New this session. Runs a 3-stage recovery sequence for a missed meeting: the same-day nudge, the reschedule offer, and the close-the-loop message if neither lands. A no-show isn't dead pipeline, it's a different message than the one you already sent. Read the skill.
The order to build them in
Start with the-list-builder and the-fit-scorer. A stale, unscored list is still the single most common reason outbound underperforms, and fixing it changes every number downstream. Run the-list-cleaner right after scoring, before anything goes into a sequence. Add the-cold-opener once you have a list worth writing to, and bring in the-call-booker right before you send, so the booking flow is ready the moment a reply says yes. The-reply-classifier, the-inbox-zero-enforcer, and the-no-show-save aren't a setup step, they're the operational layer you run every day once replies start coming in. Keep the-sequence-doctor as a recurring audit, not a one-time install: pull it out whenever reply rates move, not on a fixed schedule.
What to never hand over
Three things stay with a person, permanently, no matter how good the classifier gets. The decision to actually contact someone - these skills draft the list and the message, a human approves both. The final send, on anything: the first email, the reply, the reschedule. And any reply from a prospect that reads as angry - that gets a human response, every time, not an automated one. A desk that gets this wrong isn't more efficient, it's just faster at annoying people.
When you outgrow the manual version
These 10 skills run one job at a time, on whatever you paste in. Intempt's Outreach Rep agent is the same job description wired into your actual pipeline: it builds and scores the list from your real CDP data, drafts and sends the sequence, classifies and routes replies, and books the meeting, so you're not pasting the same inputs in every morning. Same desk. The difference is whether a person is doing the pasting or the data already lives in the system.
Frequently asked questions. Answered.
Five skills, all added this session: the-reply-classifier sorts inbound replies into Interested, Later, Referred, Objection, Dead, or Angry with evidence and a next action. The-inbox-zero-enforcer processes a full inbox into Drafted, Escalated, Scheduled, or Closed. The-list-cleaner dedupes and flags a raw list before it hits a sequence - wrong titles, stale roles, wrong companies, broken data. The-call-booker writes the booking message, the reschedule, the confirmation, and the day-before reminder. The-no-show-save runs a 3-stage recovery sequence for a missed meeting. Together they close the reply-handling gap that existed across the entire catalog before now.
