Claude skills for GTM are markdown files that teach Claude one go-to-market job. Each file says what to read first, which steps to run, what to return and where to stop. GTM teams use them because a skill returns the same shape of output every time, whoever runs it.
We publish 87 of our own skills in the Kinetyca skills directory. This page is the shortlist we would hand a GTM team first, grouped by the job each skill does. Every entry says what the skill does, when to reach for it, and links to its full page. It is one part of how we run Claude for GTM, next to the brain and the stack.
How we structure a Claude skill
Good Claude skills for GTM start with how the file is built. Our whole GTM methodology lives in markdown, more than 33,000 lines of it. Every skill our team runs is one file, and Claude Code reads that file before it starts the work.
The file is the instruction set. It names the tools Claude may open for that job and the tools it must leave alone. It says which knowledge base to read first. It also spells out what the skill is never allowed to do.
Our list build is a good example. It names the two sourcing tools and when to pick each one. It reads the client's ICP from their repo instead of inventing one. It has no direct access to the database, and it stops for a human before anything gets sent.
The filename tells you where the skill runs
Internally, every file name reads left to right. Take o4-message-1-main-copy as the example.
ois the kind of work, outbound.4is the stage, message.1is the first step inside that stage.main-copyis the job the file does.
Nobody has to open the file to know what it is or when it runs. Outbound skills run through seven stages, numbered 0 to 6, and every outbound skill sits in one of them.
| Stage | Name | What sits there |
|---|---|---|
| 0 | Setup | Credentials, sending domains and infrastructure. Nothing else runs until this is done. |
| 1 | Foundation | Who the client is, the offer, the ICP and the proof. Built once, then reused. |
| 2 | Strategy | The campaign concepts, the market sizing and the account scoring model. |
| 3 | Targeting | The list build, the email verification waterfall and the account scores. |
| 4 | Message | The main copy, the personalisation, then writing at scale. |
| 5 | Launch | The readiness check, then the campaign build in the sending tool. |
| 6 | Operate | Replies, deliverability, inbox rotation, the weekly report and the improvement loop. |
Numbers are assigned before a file is written and never reused. The public pages on our site use plain names in place of these codes, so the list below does too.
The best Claude skills for GTM, by job
These are the 15 we would start with, ranked in roughly the order a new engagement runs them. The table is the short version.
| Skill | Job | When to use it |
|---|---|---|
| 1. GTM Playbook Builder | Strategy | Before anything else, since every other skill reads from it |
| 2. Targeting Ideas | Strategy | Planning a quarter and choosing who to go after |
| 3. Prospect Researcher | Research | Before a first touch on a named account, or before a call |
| 4. Competitor Battlecards | Research | The start of a quarter, or when a competitor reprices |
| 5. Lead Sourcing Waterfall | Targeting | Building a new list from an ICP |
| 6. Email Verification Waterfall | Targeting | Before any list goes into a sequencer |
| 7. Account Scorer | Targeting | Tiering a list against a scoring model you have locked |
| 8. Post Engager Play | Signals | Running LinkedIn post engagers as a weekly channel |
| 9. Segment Copywriter | Copy | Tight segments where volume matters |
| 10. One to One Writer | Copy | Named, valuable accounts worth writing to individually |
| 11. Copy Scorer and Rewriter | Copy | An underperforming sequence, or a final check before launch |
| 12. Pre-Call Prep | Calls | Every morning, for the day's external calls |
| 13. Reply Triage | Operations | Any live campaign that is getting replies |
| 14. Daily Performance Monitor | Operations | Every live campaign, on a schedule |
| 15. The Skill That Writes Skills | Operations | A job only one person on the team does properly |
Strategy and research
1. GTM Playbook Builder
What it does. The GTM Playbook Builder takes a website URL and returns the strategy document everything else is written against. It covers a company overview, social proof, the top five verticals, ICPs with personas, and pain-to-value mapping. Verticals are ranked by the proof that already exists.
When to use it. First, for any client. Targeting ideas come from its verticals, copy quotes its proof, and battlecards position against its value props. Personalisation run without it uses a first name and says nothing.
2. Targeting Ideas
What it does. Targeting Ideas reads your playbook and returns ten campaign angles in a numbered table. Each one carries company criteria for a data tool, the personas to reach, and signal tags for timing.
When to use it. At the start of a quarter, when the hard question is who to write to. Our rule is to pick three of the ten. A team can read the table together, cut seven, and start building the same afternoon.
3. Prospect Researcher
What it does. The Prospect Researcher takes a domain or a LinkedIn profile and returns one structured snapshot. It reads the company first, then the people, then the signals: hiring, tooling, expansion and leadership changes. It closes with the two or three angles worth using, each marked by how time sensitive it is.
When to use it. Before a first touch on a named account, before a call, and during list qualification. If an account has nothing going on, the snapshot says so. It leaves the email to a copy skill, because research and copy are separate jobs.
4. Competitor Battlecards
What it does. Competitor Battlecards builds one card per competitor: positioning, strengths, weaknesses, pricing, objection handling and win-loss patterns. It compares what a competitor says about itself with what reviews and community threads say. Objection handling is written against your own playbook, proof and pricing.
When to use it. At the start of each quarter, and whenever a competitor reprices or ships something new. Batch mode runs the whole competitive set at once, so every card comes back in the same shape.
Targeting and data
5. Lead Sourcing Waterfall
What it does. The Lead Sourcing Waterfall turns an ICP into a normalised, deduplicated list. It checks data you already own first. Each provider then fills only what the last could not. Match counts come back before any credits are spent, so you approve a real number.
When to use it. Any new list, built people-direct or company-first. The company-first shape is the one an account-based motion needs. It is discovery only, and nothing is contacted.
6. Email Verification Waterfall
What it does. The Email Verification Waterfall verifies the addresses you have. When one fails, it tries four finders in sequence and verifies each candidate by the same rule. Catch-all domains get a second check of a different kind, since a standard pass on a catch-all means very little.
When to use it. Before any list reaches a sequencer. Bounces damage the sending reputation the rest of the list depends on. The skill never sends anything, including a test, and people with no findable address come back marked as exactly that.
7. Account Scorer
What it does. The Account Scorer applies a locked scoring model to a real list. Disqualifiers are applied first. Fit and signal are scored separately, and every account gets a tier with one line explaining it. A dimension the data cannot fill is removed and the weights re-normalised, so missing data is never scored as a zero.
When to use it. Once your scoring model is defined and calibrated. The skill applies the model and never re-tunes it mid-run, so the same list and model always give the same tiers.
8. Post Engager Play
What it does. The Post Engager Play chains the whole engager motion. It harvests the people who reacted to chosen posts, then enriches, qualifies against your ICP, verifies and writes. Qualified, verified leads land on your standing campaigns as drafts, each with a personalised sequence.
When to use it. When you want LinkedIn engagement as a standing weekly channel. Chained, the engagement is still recent when the copy is written. Activation stays a separate human decision.
Claude skills for sales: copy and calls
9. Segment Copywriter
What it does. The Segment Copywriter writes one three-email sequence for a whole segment. Emails 1 and 2 share a thread, and email 3 opens a fresh thread as the breakup. Subject lines stay at three words, lowercase.
When to use it. When the segment is tight enough that the list carries the relevance, and volume matters. It deliberately does no company-level research. Change the segment, rerun it, and compare against last week's numbers.
10. One to One Writer
What it does. The One to One Writer writes a three-email sequence per named person. It picks one of five frameworks for each person. In list mode it returns the same file with the bodies and subject lines written into new columns.
When to use it. When the accounts are named, few and valuable enough to write to individually. Offers, proof and tone all come from your playbook, so nothing invented reaches a prospect. For tight segments, the Segment Copywriter is usually the better fit.
11. Copy Scorer and Rewriter
What it does. The Copy Scorer and Rewriter scores a sequence across eight dimensions and names the lines that cost it points. It fixes what fails, keeps what works, then scores the rewrite again. Two of its checks are whether every claim is real and permitted, and whether the touches repeat each other.
When to use it. On a campaign with a suspected copy problem, on inherited copy, and as a check before launch.
12. Pre-Call Prep
What it does. Pre-Call Prep starts with your own history: every email, the source campaign, notes, tags and open objections. Then it adds fresh web research. Give it a name for a full brief, or nothing for a table of every external call on today's calendar.
When to use it. Once every morning. Prep usually fails on timing, since nobody digs through a CRM five minutes before a call. One run covers the whole day. After the call, the Post-Call Debrief turns the transcript into action items and a ready to send follow up.
Operations
13. Reply Triage
What it does. Reply Triage sorts every inbound reply into nine categories, from super interested to wrong person to unsubscribe. Opt-outs are suppressed across every campaign. Out-of-office contacts are rescheduled for their return, and the next message is drafted for a person to approve.
When to use it. On any live campaign that is getting replies. It runs continuously, so drafts for the highest-intent replies are ready while the prospect is still in the thread. Inbound text is treated as content to read, never as instructions to follow.
14. Daily Performance Monitor
What it does. The Daily Performance Monitor posts a kill, stay or scale verdict on every live campaign each morning. It evaluates each sending split on its own and reads deliverability alongside reply rate.
When to use it. Always, on a schedule. Stay is the default, and a kill needs a campaign that is dead or consistently poor over a real window. It only reports. Acting on the verdict is a separate skill with its own confirmation step.
15. The Skill That Writes Skills
What it does. The Skill That Writes Skills turns a plain process description into a finished skill file. It interviews you first about the edge cases and decisions the description left out. You get the file, five starter prompts and a troubleshooting guide.
When to use it. When a job lives in one person's head, which makes that person a bottleneck and a risk. A skill file gives everyone the same output shape, so the work becomes reviewable and comparable week to week.
What good Claude skills for GTM have in common
Read the list again and one pattern shows up. The Claude skills for GTM that hold up in production stop before the irreversible step. Lead sourcing is discovery only. The engager play stages drafts. Reply triage drafts and waits. The monitor reports and changes nothing.
We make that choice in every file. Claude does the research, the sorting, the scoring and the drafting. A person still approves the spend, activates the campaign and sends the first reply.
- 01Build the playbookRun the GTM Playbook Builder first. Everything else reads from it.
- 02Build and verify one listSource it through the waterfall, then verify every address.
- 03Write, score, then operateWrite the copy, score it before launch, then add Reply Triage and the monitor.
For a job we have not published, the guide below gives you the skill file that writes skill files.
When you say you have an automated system and a brain, what does that actually look like?
For every client we create a separate GitHub repository. It holds the intelligence we gather from the start: sales calls, CRM information, website visitors, and content and ad metrics. The skills read from that repository, so the method is shared and the knowledge stays with the client.
If we are already using Claude, how do skills like these fit in?
Claude skills for GTM are markdown files, so they sit inside the Claude setup you already have. One person can start with a handful of skills in a single Claude project. Once several people share them, the skills move into a shared GitHub repo, and MCPs connect them to the live tools. Each skill page lists the input it expects.
Can we run the tools and skills ourselves later?
Yes. Clients often start with us running the system and later bring it in-house. We hand over the documentation, recorded walkthroughs and training. Sometimes we also help hire the person who will own it.
Is it AI agents doing the outreach? How much is automated?
Claude does the research, sourcing, scoring, drafting and monitoring. The irreversible steps stay with a person. Campaigns are built as drafts, replies are drafted for approval, and a human activates every launch.
Software, accelerator and startup background. Runs the GTM architecture side: how the system is designed, what it is built on, and where AI actually earns its place.
LinkedIn ↗YouTube ↗