How to use the ZoomInfo Sales Lookup Guide agent skill

What the skill makes the agent do

The skill applies only when a ZoomInfo connector is present and connected to the agent. It does not connect to ZoomInfo by itself. Its job is to choose the right lane: search to find candidates, enrich for verified fields, research for narrative briefs, or similarity and recommendation lookups for "more like this" requests.

It also enforces discipline. The agent looks up standard category values before searching, resolves the exact company or person before using IDs, checks that results meet your filters, and says what was not found instead of guessing.

Install the skill from the ZIP

Download the ZIP from this page and unzip it. For a personal setup, place the unzipped folder in ~/.claude/skills/. For a single project, place it in the project's .claude/skills/ folder. Alternatively, in the Claude apps, open Settings, go to Capabilities, and upload the skill under Skills.

The package contains two files: SKILL.md, which holds the instructions, and agents/openai.yaml. The skill does not run scripts.

Trigger the skill with natural requests

The skill loads when a focused Sales workflow has selected ZoomInfo, or when you explicitly ask for ZoomInfo company search, contact search, enrichment, intent, similarity, recommendations, or research. Name ZoomInfo in your request and give the constraints you care about.

If you use a different enrichment source in the workflow, the skill will not apply.

• Find 50 SaaS companies in the UK with 200 to 1,000 employees using ZoomInfo.
• Use ZoomInfo to enrich these three domains with verified contact details.
• Prepare a ZoomInfo account research brief on this company before my call.
• Find companies similar to this ZoomInfo company ID.

Know which action the agent picks

The skill separates ZoomInfo actions into lanes so the agent does not blur them. Discovery actions include lookup, search_companies, search_contacts and search_intent. Structured enrichment uses enrich_companies, enrich_contacts and enrich_intent. Narrative research uses account_research and contact_research. Similarity and recommendation use find_similar_companies, find_similar_contacts and get_recommended_contacts. submit_feedback is used only when you ask for it.

• Search first to build a shortlist, then enrich the shortlist.
• Use research when you want a brief, not a table.
• Use similarity for expansion, not as a replacement for precise filters.

Understand what the skill needs and its safety notes

You need a connected and present ZoomInfo connector exposed to the agent, the MiMo Sales parent workflow or an equivalent that selects this skill, and ZoomInfo account and API access. Enrichment and research use credits, so the skill asks for the smallest credit-consuming step that answers your question, but usage still costs money.

The skill runs no scripts, but it uses the network because it calls ZoomInfo through the connector. The guide says ZoomInfo output is not CRM ground truth. Ownership, deal status and customer status should come from your CRM, and relationship or engagement context from ZoomInfo is labelled as connector-surfaced.

• Check your credit balance before large enrichment or research runs.
• Treat connector-surfaced relationship context as unverified until checked elsewhere.

Tips for better results

Give the agent strong identifiers. A domain, company ID, business email or exact name plus company resolves faster and more reliably than a name alone. Give hard filters such as geography, headcount and seniority up front, because the agent verifies results against them and will report near matches separately.

Expect bounded results. The guide does not claim exhaustive market coverage from one result page, so ask for more pages or narrower filters if you need completeness.