Agent skill
icp-personas
Define the ICP, size and tier an ABM target-account list, and map the buying committee — the strategy layer before list-building and enrichment.
Filed under Prospecting and list building and Positioning and messaging.
From Cold-IQ/coldiq-marketplace-skills · 17 skill entries · 1 · pushed 2026-09-22
What it does when it runs
Define the ICP, size and tier an ABM target-account list, and map the buying committee — the strategy layer before list-building and enrichment. Use when defining an Ideal Customer Profile and scoring criteria, deciding how many accounts to target for a revenue goal, tiering accounts (A/B/C/D), mapping personas/buying committee (champion, economic buyer, end user, technical evaluator), or planning persona-based messaging. Triggers on "ICP", "ideal customer profile", "who to target", "scoring criteria", "ABM accounts", "how many accounts", "account tiering", "buying committee", "persona mapping", "champion", "economic buyer", "revenue reverse-engineering". Do NOT use for actually sourcing companies/contacts (see apollo-search / coldiq-search-enrich), scoring math on an export (see tam-scoring), or writing copy (see crawford-method / cold-email-copy).
Automated analysis of the skill and the 0 files bundled beside it. A skill’s own description is written to be selected by an agent, so it describes the job and not the dependencies.
- Keys and connectors you must supply
- None found.
- Hosts it reaches
- No third-party host appears in the skill or its bundled files.
- Tool permissions it declares
- No
allowed-toolsin the frontmatter. It only issues instructions, so there is nothing to bound. - Actions present in the files
- None. Instructions only.
Install it
View source on GitHub ↗git clone --depth 1 --filter=blob:none --sparse https://github.com/Cold-IQ/coldiq-marketplace-skills.git /tmp/coldiq-marketplace-skills git -C /tmp/coldiq-marketplace-skills sparse-checkout set "skills/icp-personas" mkdir -p ~/.claude/skills/icp-personas cp -R "/tmp/coldiq-marketplace-skills/skills/icp-personas/." ~/.claude/skills/icp-personas/
Picked up without a restart. A project skill of the same name is shadowed by your personal one. For one repository only, swap ~/.claude/skills for .claude/skills. Claude Code docs ↗
Or take the whole library
This repo ships a .claude-plugin manifest, so Claude Code can install all 17 skills at once. Plugin skills are invoked as /<plugin>:<skill>, so they never collide with your own.
/plugin marketplace add Cold-IQ/coldiq-marketplace-skills /plugin
The folder is the same in every client that implements the format — 46 of them — so if yours is not above, only the destination changes.
The skill
Source on GitHub ↗Reproduced in full from Cold-IQ/coldiq-marketplace-skills/blob/495c1b4256db8c6683ac1ff59a093899228eca73/skills/icp-personas/SKILL.md, which is licensed MIT (repository). 522 words, 6 headings.
ICP, Accounts & Personas
The strategy layer: decide WHO to target, HOW MANY, and WHICH people inside each account — before you spend credits sourcing and enriching. Mostly methodology; a few steps source data from ColdIQ.
ColdIQ Marketplace Endpoints (where strategy touches data)
| Task | Method | Path | Credits | Endpoint ID | Notes |
|---|---|---|---|---|---|
| Validate filter values (ICP criteria) | POST | /v1/limadata/references/autocomplete | free | limadata.references.autocomplete | Confirm titles/industries exist |
| Size the addressable market | POST | /v1/ai-ark/companies | per result | ai_ark.companies.search | How many fit the ICP |
| Layer intent for prioritization | GET | /v1/signalbase/funding-signals | unknown | signalbase.funding_signals | Bump in-market accounts |
| Map the buying committee | POST | /v1/limadata/prospect/employees | 25 | limadata.prospect.employees | Find personas at an account |
1. Define the ICP (3 layers + scoring)
- Firmographic: industry, employee size, revenue, geography, growth, funding stage.
- Technographic: tech stack, CRM, marketing automation, competitor tools.
- Behavioral/intent: hiring, funding, leadership changes, content engagement, website visits.
Reverse-engineer from your best 10–20 closed-won customers. Score 100 pts (example weights): industry 20 · size 15 · revenue 15 · geography 10 · tech fit 15 · growth 10 · intent 15. Tiers: A 90–100 (1:1 ABM) · B 70–89 (1:few) · C 50–69 (programmatic) · D <50 (exclude).
Validate your criteria translate to real filter values before sizing:
→ POST /v1/limadata/references/autocomplete · free · limadata.references.autocomplete
(Hand the scoring rules to tam-scoring to run on an actual export.)
2. Size & select the account list (ABM)
Revenue reverse-engineering — work backward from the target through funnel benchmarks: Identified→Aware 55% · Aware→Interested 32% · Interested→Considering 18%. (e.g. $1M ARR target ≈ ~3,367 accounts identified.)
Check how many companies actually fit before committing:
→ POST /v1/ai-ark/companies · per result · ai_ark.companies.search
4-layer selection: firmographic fit · technographic indicators · CRM intelligence (closed-lost,
lost-to-competitor, churned) · lookalike modeling from best customers. Then prioritize the
in-market ones:
→ GET /v1/signalbase/funding-signals · ? cr · signalbase.funding_signals (unverified)
Track stage progression: Identified → Aware → Interested (5+ clicks / 10+ engagements) → Considering (site visits, content, demo interest).
3. Map the buying committee
| Role | Function | Budget priority |
|---|---|---|
| Champion | Internal advocate driving evaluation | 40–50% |
| Economic Buyer | Signs the check; cares about ROI | 20–30% |
| End User | Daily user; cares about UX/workflow | 15–20% |
| Technical Evaluator | Integration, security, compliance | 5–10% |
| Blocker/Gatekeeper | Can veto, rarely initiates | monitor |
For each persona capture: title patterns + seniority, function, jobs-to-be-done, pains, success
metrics, content preference, buying role. Find the actual people once you've defined the personas:
→ POST /v1/limadata/prospect/employees · 25 cr · limadata.prospect.employees
Messaging matrix: per persona, different JTBD + pain + stage-appropriate content + role-appropriate CTA (champion → demo/ROI, end user → trial/ease-of-use). Hand off to crawford-method / cold-email-copy to write it.
Where this hands off
ICP scoring rules → tam-scoring · sourcing companies/contacts → apollo-search / coldiq-search-enrich · personas-into-ads → ad-audiences · the full build order → campaign-delivery.
Other skills for the same job
Different authors, same problem. Matched on the words in the skill name, across every library in the catalogue except this one.
- icp-research-assistant by irinabuht12-oss · 3,698
- sales-icp by zubair-trabzada · 1,407
- icp-onboarding by growthenginenowoslawski · 739
- icp-prompt-builder by growthenginenowoslawski · 739
- icp-builder by OpenClaudia · 708
- reddit-icp-monitor by Varnan-Tech · 672
- icp-import by Othmane-Khadri · 317
- icp-definer by Othmane-Khadri · 317
Need help setting it up?
This page tells you what icp-personas does and what it needs. Cheetah builds the agent setup it runs inside: data, CRM, sequencing and the guardrails.
Book a call →The directory stays free. There is nothing gated behind this.