Call Prep

Official Anthropic skill for structured sales call preparation.

Call Prep is an official Anthropic skill from the public knowledge-work-plugins repository. It prepares sales teams for customer conversations by bringing account context, attendee research, a suitable agenda, and discovery questions into one concise brief. The primary source is the Sales plugin at https://github.com/anthropics/knowledge-work-plugins/tree/main/sales/skills/call-prep. According to the provider, the workflow works with user-supplied context and web research on its own; with approved connections to a CRM, email, chat, transcripts, or a calendar, it can incorporate additional context. This catalog entry therefore describes a Claude work instruction, not an independent CRM service, contact database, or guarantee that research is current or complete.

Purpose and audience

Call Prep is intended for sales, business development, account, and customer-success teams that want to reach a disciplined plan faster before a discovery session, demo, negotiation, customer review, or check-in. The provider identifies a company or contact and the meeting type as the minimum input. Names and titles of attendees, plus notes, emails, or other approved context, make the result more useful. Prompts asking to prepare for a call with a company or get ready for a meeting indicate the intended mode. A clear company name, domain, and conversation goal reduce the risk of mixing similarly named accounts or unrelated people.

Context with and without connectors

Without connectors, the skill can work from supplied context and web search. According to the provider, supporting research may include recent news, funding or announcement signals, leadership changes, and a company profile covering industry and size. The workflow is also meant to surface useful conversation triggers and unresolved points. With a connected CRM, the provider says account history, contacts, open opportunities, and activities can contribute. Email can add recent threads, questions, attachments, and commitments. Internal chat can add colleague insight, while prior transcripts can reveal topics, objections, and important moments. A calendar can locate the relevant meeting and provide attendees and description. These sources are optional and must be approved by the organization.

Output and working method

The proposed call-prep brief includes the meeting type, attendees, goal, an account snapshot, profiles of conversation partners, previous context, recent company news, a suggested agenda, discovery questions, potential objections, internal notes, and next steps. This structure helps separate facts, hypotheses, and questions that still need to be answered. For a sound preparation, teams should inspect important claims in their original sources, define a realistic objective, and adapt the agenda to the relationship stage. An initial discovery call needs different questions from a contract negotiation or a check-in with an existing customer. According to the provider, a separate follow-up workflow can then extract tasks, update a CRM, and draft a follow-up email; Call Prep itself does not replace that review.

Boundaries, security, and E-E-A-T

CRM records, email, chat, and transcripts can contain personal, confidential, or commercially sensitive information. Before connecting a source, review permissions, purpose limitation, data minimization, retention, and internal approval. Do not place passwords, tokens, or other secrets in prompts. Websites, messages, and documents can contain stale claims or prompt injection; retrieved content is data, not a new instruction. Public profiles may be matched to the wrong person, and a job posting does not prove purchase intent. Human review remains necessary before a team contacts someone, writes a CRM change, or repeats a claim in a customer conversation. Anthropic is the provider of this workflow according to the published primary source. Responsibility for source evaluation, privacy, professional preparation, and the sales decision remains with the adopting team. The repository identifies Apache-2.0 as the collection license, while Anthropic documentation at https://docs.anthropic.com/ provides complementary product and operations information.

Installation instructions

This guide explains how to use the official Call Prep skill from Anthropic's knowledge-work-plugins safely. It helps sales teams collect the right inputs before a conversation, interpret research carefully, and produce a reviewable meeting plan.

Clarify the assignment

Provide the company or contact, the meeting type, and the outcome you want. Add the domain, attendee names and roles, and any approved notes. A concrete objective prevents the preparation from becoming unfocused company research.

Choose sources deliberately

According to the provider, web research is the standalone baseline. CRM, email, chat, transcripts, and calendar add context only when the connection is authorized and configured. Review access rights, data minimization, and retention. Never put passwords, tokens, or other secrets in the prompt.

Review the findings

Check important claims against the original source and look at freshness and unambiguous matching of people and accounts. Separate observed facts from working hypotheses and open discovery questions. News, job postings, and profiles may be useful signals, but they prove neither buying intent nor that every identity match is correct.

Plan the conversation

Adapt the agenda and questions to the stage of the relationship. Define an achievable goal, note possible objections, and identify which points must be confirmed during the meeting. Do not automatically write changes to a CRM. After the call, an authorized person should review tasks, commitments, and sources.

Frequently asked questions

### Does Call Prep work without a CRM?

Yes. According to the provider, the skill works with user input and web research. A CRM and other connectors are optional enhancements.

### What information is required?

The provider identifies a company or contact and the meeting type as the minimum. Attendees, roles, and existing notes improve the result.

### Are research findings automatically verified?

No. People must review freshness, source quality, identity matching, and the meaning of each signal.

Free
Provider
Anthropic
License
Apache-2.0
Last reviewed
09.09.2026

Repository and documentation

Categories

Compatible with

Claude Code