gtm-pipeline:setup
Skill1 starsChanged 5 months agoMITView on GitHub
Onboarding skill — walks a new user through installation, API key configuration, PhantomBuster agent IDs, first client setup (auto-research from domain + 5 refinement questions), and a skill tour. Run this first after cloning the repo. Also triggers on "set up gtm skills", "configure gtm", "gtm onboarding", "first time setup".
- Reads credentials
1 other project uses a version of this, 1 of them edited.
Adopted by
Most recently taken up by
- keinsaasforever/gtm-pipeline-skillsedited, first seen Oct 13, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 1
- keinsaasforever/gtm-pipeline-skills★ 7585% same
Wrote their own version · 0
None found.
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository23 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- article-writingSkill
- brand-voiceSkill
- deep-researchSkill
- email-personalizationSkill
- exa-searchSkill
- gtm-pipeline:company-enrichmentSkill
- gtm-pipeline:company-searchSkill
- gtm-pipeline:contact-filterSkill
- gtm-pipeline:demoSkill
- gtm-pipeline:outreachSkill
- gtm-pipeline:people-enrichmentSkill
- gtm-pipeline:people-searchSkill
- gtm-pipeline:pipelineSkill
- gtm-pipeline:signal-searchSkill
- investor-materialsSkill
- investor-outreachSkill
- lead-intelligenceSkill
- marketing-campaignSkill
- market-researchSkill
- research-opsSkill
- social-graph-rankerSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
19 rules. About 0 are common, 7 are distinctive, and 12 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
12
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 41
Check whether an env file exists and which keys are already set.
- Very specific: line 45
Check .agents/skills/_shared/local.md for GTM_ENV_PATH
- Very specific: line 46
Default location if not set: ~/.env.gtm
- Very specific: line 47
If neither exists: offer to create ~/.env.gtm now
- Very specific: line 66
Optional: If user won't use PhantomBuster (no LinkedIn automation), skip PHANTOMBUSTER_API_KEY, LINKEDIN_SESSION_COOKIE.
- Distinctive: line 74
If the env file doesn't exist yet:
- Distinctive: line 93
Only needed if user will use PhantomBuster for LinkedIn automation.
- Very specific: line 97
If the PhantomBuster MCP is available, use PHANTOMBUSTER_GET_AGENTS_FETCH_ALL to list all agents.
- Very specific: line 114
Write all matched IDs to _shared/local.md:
- Very specific: line 127
If PB MCP is not available, show the user _shared/local.example.md and instruct them to:
- Distinctive: line 134
If user doesn't plan to use PhantomBuster, skip this step.
- Distinctive: line 140
Ask: "Do you want to set up your first client now?
- Distinctive: line 142
If user declines, skip to Step 5.
- Very specific: line 146
Ask for a client slug (e.g.
- Distinctive: line 159
Before asking any questions, research the client's domain automatically.
- Very specific: line 162
Use firecrawl_scrape or firecrawl_crawl on the domain.
- Very specific: line 171
Search: "{company name}" OR site:{domain} with recent date filter.
- Very specific: line 181
Do not wait for user input.
- Very specific: line 273
After collecting answers, update context/profile.md and context/icp.md with confirmed values.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
GTM Pipeline Setup
Welcome. This skill walks you through everything needed to start using the GTM pipeline skills — from installation to your first client ICP.
Read .agents/skills/_shared/conventions.md before executing.
When to Use
- First time after cloning the repo
- Adding a new API key or PhantomBuster agent
- Setting up a new client (use Step 4 only)
- Troubleshooting a missing config value
Step 1 — Verify Installation
The skills live in this repo at .agents/skills/ and are auto-discovered by any agent session opened at the repo root — no install step.
ls .agents/skills/ | grep gtmExpected output: 10 directories — gtm-setup, gtm-company-search, gtm-company-enrichment, gtm-contact-filter, gtm-demo, gtm-outreach, gtm-people-enrichment, gtm-people-search, gtm-pipeline, gtm-signal-search.
If any are missing: re-clone or restore the repo — the skills ship with it.
Confirm with the user before continuing.
Step 2 — Environment Variables
Check whether an env file exists and which keys are already set.
Where to look
- Check
.agents/skills/_shared/local.mdforGTM_ENV_PATH - Default location if not set:
~/.env.gtm - If neither exists: offer to create
~/.env.gtmnow
Required keys
Walk through each key. For each one, check if it's already set in the env file. Only prompt for missing ones.
| Key | Provider | Required For | Sign Up |
PIPE0_API_KEY | Pipe0 | People search, company search | pipe0.com |
FULLENRICH_API_KEY | FullEnrich | Email enrichment (waterfall) | fullenrich.com |
BETTERCONTACT_API_KEY | BetterContact | Phone enrichment (optional) | bettercontact.rocks |
SERPAPI_API_KEY | SerpAPI | Domain lookup, web search | serpapi.com |
PARALLEL_API_KEY | Parallel AI | FindAll, task enrichment | parallel.ai |
APIFY_API_KEY | Apify | SimilarWeb traffic data | apify.com |
FIRECRAWL_API_KEY | Firecrawl | Website crawl, signal search | firecrawl.dev |
PHANTOMBUSTER_API_KEY | PhantomBuster | LinkedIn automation | phantombuster.com |
LINKEDIN_SESSION_COOKIE | PhantomBuster auth | Your browser (see below) | |
OPENROUTER_API_KEY | OpenRouter | Signal assessment LLM | openrouter.ai |
Optional: If user won't use PhantomBuster (no LinkedIn automation), skip PHANTOMBUSTER_API_KEY, LINKEDIN_SESSION_COOKIE.
For LINKEDIN_SESSION_COOKIE: Tell the user to open LinkedIn in Chrome, go to DevTools → Application → Cookies → li_at value.
Minimum viable set (to run a basic demo): PIPE0_API_KEY, FULLENRICH_API_KEY, SERPAPI_API_KEY, PARALLEL_API_KEY, FIRECRAWL_API_KEY, OPENROUTER_API_KEY.
Writing keys
If the env file doesn't exist yet:
touch ~/.env.gtmFor each missing key, append it:
echo 'KEY_NAME=value' >> ~/.env.gtmThen set GTM_ENV_PATH in _shared/local.md:
GTM_ENV_PATH=/Users/{username}/.env.gtmStep 3 — PhantomBuster Agent IDs
Only needed if user will use PhantomBuster for LinkedIn automation.
Option A — Auto-lookup via PhantomBuster MCP
If the PhantomBuster MCP is available, use PHANTOMBUSTER_GET_AGENTS_FETCH_ALL to list all agents. Match agent names to config keys:
| Config Key | Agent Script Name |
PB_AGENT_CONNECT | LinkedIn Auto Connect |
PB_AGENT_MESSAGE | LinkedIn Message Sender / Sales Navigator Message Sender |
PB_AGENT_SN_ACCOUNT | Sales Navigator Account Scraper |
PB_AGENT_EMPLOYEES | LinkedIn Company Employee Finder / LinkedIn Search Export |
PB_AGENT_SN_SEARCH | Sales Navigator Search Export |
PB_AGENT_PROFILE | LinkedIn Profile Scraper |
PB_AGENT_JOB_SEARCH | LinkedIn Search Export (jobs mode) |
PB_AGENT_LIKER | LinkedIn Auto Liker |
PB_AGENT_COMMENTER | LinkedIn Auto Commenter |
PB_AGENT_WITHDRAWER | LinkedIn Invitation Auto Withdraw |
PB_AGENT_INBOX | LinkedIn Inbox Scraper |
PB_AGENT_CONNECTIONS | LinkedIn Connections Export |
Write all matched IDs to _shared/local.md:
## PhantomBuster Agent IDs
| Config Key | Agent ID |
|------------|----------|
| PB_AGENT_CONNECT | 123456789 |
| PB_AGENT_SN_ACCOUNT | 987654321 |
...Option B — Manual
If PB MCP is not available, show the user _shared/local.example.md and instruct them to:
- Log in to PhantomBuster dashboard
- Find each agent and copy its ID from the URL or settings panel
- Fill in
_shared/local.md(copy fromlocal.example.md)
Skip entirely
If user doesn't plan to use PhantomBuster, skip this step.
Step 4 — First Client Setup
Ask: "Do you want to set up your first client now? If yes, provide the client's domain (e.g. acme.com)."
If user declines, skip to Step 5.
4.1 Create client directory
Ask for a client slug (e.g. acme for acme.com). Create the working directory:
{slug}-gtm/
├── context/
├── prompts/
└── csv/
├── input/
├── intermediate/
└── output/4.2 Auto-research the domain
Before asking any questions, research the client's domain automatically. Run three sources in parallel:
Firecrawl — scrape homepage + about/careers pages:
Use firecrawl_scrape or firecrawl_crawl on the domain. Extract:
- What they sell / core product or service
- Target customer (explicit or implied)
- Industries served
- Key value propositions
- Company size / funding stage signals
- Tone of voice from copy
SerpAPI — recent news and announcements:
Search: "{company name}" OR site:{domain} with recent date filter. Extract:
- Funding rounds
- Company description from third-party sources
- Customer segments mentioned in press
Parallel web search (if Parallel API key set):
Query: "{company}" {domain} — what they sell, who they target, key customers, industries served
4.3 Draft profile.md and icp.md immediately
Do not wait for user input. Draft both files from research findings now. Mark inferred fields with [INFERRED].
Draft context/profile.md:
# {Company Name} — Client Profile
## What They Sell
{product/service description from research}
## Target Buyer
{who buys from them — role, company type, industry}
## Value Proposition
{how they describe their own value}
## Reference Customers / Segments
{any named customers or segments found}
## Tone of Voice
{formal/casual, technical/plain — inferred from website copy}
## Notes
- Profile auto-researched from {domain} on {date}
- Fields marked [INFERRED] should be confirmed with the clientDraft context/icp.md:
# {Company Name} — ICP Definition
## Company ICP
### Target Industry
{industries from research} [INFERRED]
### Company Size
- Min employees: [INFERRED — ask to confirm]
- Max employees: [INFERRED — ask to confirm]
### Location
{geography signals from research} [INFERRED]
### Revenue Range
[INFERRED — ask to confirm]
## Contact ICP
### Target Job Titles / Roles
{roles from research} [INFERRED]
### Seniority
[INFERRED]
## Contact Filter Configuration
### Job Tier Definitions
[To be filled after Step 5 refinement questions]
### Industry Tiers
[To be filled after Step 5 refinement questions]
### Location Tiers
[To be filled after Step 5 refinement questions]
### Company Size Filter
[To be filled after Step 5 refinement questions]
### Sort Order
job_tier → industry_tier → comp_loc_tier → person_loc_tier → employee_count_missing → icp_kw_score (desc)Show the user what was auto-filled vs. inferred, then ask 5 refinement questions:
4.4 Five Refinement Questions
- Who are your hardest-to-replace customers? (This defines your real ICP — not who you want to sell to, but who gets the most value and keeps paying)
- Who should we NOT target? (Anti-ICP) (Industries, company types, roles, or geographies that are a bad fit, take too long to close, or drain support)
- Which job titles actually convert? (Not who answers the phone — who signs the contract or champions the deal internally)
- What's your geographic priority? (Which countries/regions come first? Are there any we should skip entirely?)
- What tone do you use in outreach? (Formal or casual? Any message examples you love or hate? Languages?)
4.5 Update files with refinement answers
After collecting answers, update context/profile.md and context/icp.md with confirmed values. Remove [INFERRED] markers where confirmed. Fill in the Contact Filter Configuration section in icp.md.
Step 5 — Skill Tour
Show available skills with a one-line summary:
## GTM Pipeline Skills
/gtm-pipeline:setup You are here — onboarding and first client config
/gtm-pipeline:pipeline Orchestrator — plan and run the full pipeline end-to-end
/gtm-pipeline:company-search Build a company list from Sales Navigator, Parallel FindAll,
Firecrawl Agent, or web scraping
/gtm-pipeline:company-enrichment Enrich companies with headcount, revenue, growth metrics.
Run ICP scoring (0–100) against your icp.md
/gtm-pipeline:signal-search Find buying intent signals (funding, hiring, leadership changes).
Scores each signal 1–100 for purchase intent
/gtm-pipeline:people-search Find contacts at target companies by job title, or discover
people by persona (no company list needed)
/gtm-pipeline:contact-filter Classify and rank contacts before enrichment — saves credits
by rejecting non-ICP contacts early
/gtm-pipeline:people-enrichment Enrich contacts with verified work email and phone number
/gtm-pipeline:outreach Run LinkedIn connection requests and personalized messages
via PhantomBuster — one profile at a time
/gtm-pipeline:demo Generate a demo lead list (~10 enriched contacts + messages)
from an ICP description — no company list neededSuggested first run:
/gtm-pipeline:demoRuns a mini end-to-end pipeline for 10 contacts. Good way to verify keys are working and see the output format before committing credits to a full run.
For a full pipeline:
/gtm-pipeline:pipelineAssesses your starting data, picks Company-First or Signal-First workflow, estimates costs, and walks you through execution.
Configuration Reference
Files written by this skill
| File | Purpose |
~/.env.gtm (or $GTM_ENV_PATH) | API keys |
.agents/skills/_shared/local.md | PB agent IDs, env path |
{slug}-gtm/context/profile.md | Client profile (what they sell, tone) |
{slug}-gtm/context/icp.md | ICP definition (job tiers, industries, locations, size) |
Re-running setup
This skill is safe to re-run at any time:
- It only appends missing keys — won't overwrite existing ones
- It will update local.md agent IDs if PB MCP is available
- It will ask before overwriting existing profile.md or icp.md
What's Missing (To Document)
- Automated key validation (test API calls to confirm each key works before writing)
- Multi-client batch setup (configure several clients at once)
- Upgrade path: re-run setup to add new skills after pulling from the repo
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

