Run any Skill in Manus with one click

meeting-prep

Stars3

Forks0

UpdatedJune 8, 2026 at 10:25

Build a full client profile before a meeting. Give a name + company, get an 8-part profile with cited sources. An example of an Act-layer agent.

Installation

Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.

Run Skill in Manus

Source

automation-flow

automation-flow/cba-starter

View GitHub Repository View Creator Repositories

Download

Run Skill in Manus

Related occupationsSOC

Based on SOC occupation classification

Market Research Analysts and Marketing SpecialistsBusiness and Financial Operations Occupations·SOC 13-1161

SKILL.md

readonly

name	meeting-prep
description	Build a full client profile before a meeting. Give a name + company, get an 8-part profile with cited sources. An example of an Act-layer agent.

Meeting-Prep

An example agent, so you can see what one looks like. Give it a person's name and their company, get a meeting-ready profile instead of an hour of prep. Every claim has a source.

Run it

Input: a person's name + their company.

Read this brain first. Check wiki/ for any page on the person, the company, or past meetings, and fold what's there into the profile.
Validate on LinkedIn that the person works there now, in that role. If you can't confirm it, say so. Don't guess.
Research the company live: what they do, products, stack, and news from the last 12 months. Cite a source for every claim.
Read the last 60 days of the person's posts: recurring themes, tone, and 2-3 ice breakers to open with.
Assemble the profile, 8 parts:
- Key contacts: name, role, place in the decision
- Company: site, industry, size, products
- Systems: CRM/ERP, tools, integration constraints
- Challenges and pains
- Goals: short term and long term
- Budget and timeline
- Decision structure: who signs off
- Research and talking points: findings plus ice breakers

Rules

Every claim has a cited source.
Anything you can't verify in public: mark "to verify in the discovery call". Never invent facts, numbers, or quotes.
Save the result as wiki/person/.md, linked with [[wikilinks]] to the company.
If the name is in Hebrew, write the profile in Hebrew.

This is the shape of an Act-layer agent: it reads the Brain, does focused work, and writes back to the Brain. Copy it to build your own.

Want the production-grade version, with the Word template and the full research protocol? → https://github.com/automation-flow/meeting-prep

More from this repository

same repository

connect-clickup

automation-flow/cba-starter

Connect ClickUp tasks to the brain with a personal API token and a pull script, so open work lands in raw/. Use when the user manages tasks or projects in ClickUp.

2026-06-083

connect-github

automation-flow/cba-starter

Connect GitHub to the brain with the gh CLI, so pull-github.sh pulls repo activity (commits, issues) into raw/. Use when the user wants their code or repo activity captured.

2026-06-083

connect-gmail

automation-flow/cba-starter

Connect Gmail to the brain — the easy connector way (browser sign-in, no key) or as a saved local copy via the gws CLI. Use when the user wants their email readable by the brain.

2026-06-083

connect-gws

automation-flow/cba-starter

Install the Google Workspace CLI (gws) and sign the user in, so the brain can pull Google Drive, Gmail, and Calendar into raw/ as local copies. Use when connecting any Google source as a saved local copy rather than a live connector.

2026-06-083

connect-schedule

automation-flow/cba-starter

Wire the brain's scripts to run on a schedule — pull, ingest, lint, backup — so it stays current on its own. Use after at least one source is connected and the user wants hands-off refresh.

2026-06-083

connect

automation-flow/cba-starter

Turn on backup and wire the brain to the user's tools, with them, one at a time. Run after /onboard, or any time to add a source. Sync first, then sources, then an optional schedule.

2026-06-083

name	meeting-prep
description	Build a full client profile before a meeting. Give a name + company, get an 8-part profile with cited sources. An example of an Act-layer agent.

Meeting-Prep

An example agent, so you can see what one looks like. Give it a person's name and their company, get a meeting-ready profile instead of an hour of prep. Every claim has a source.

Run it

Input: a person's name + their company.

Read this brain first. Check wiki/ for any page on the person, the company, or past meetings, and fold what's there into the profile.
Validate on LinkedIn that the person works there now, in that role. If you can't confirm it, say so. Don't guess.
Research the company live: what they do, products, stack, and news from the last 12 months. Cite a source for every claim.
Read the last 60 days of the person's posts: recurring themes, tone, and 2-3 ice breakers to open with.
Assemble the profile, 8 parts:
- Key contacts: name, role, place in the decision
- Company: site, industry, size, products
- Systems: CRM/ERP, tools, integration constraints
- Challenges and pains
- Goals: short term and long term
- Budget and timeline
- Decision structure: who signs off
- Research and talking points: findings plus ice breakers

Rules

Every claim has a cited source.
Anything you can't verify in public: mark "to verify in the discovery call". Never invent facts, numbers, or quotes.
Save the result as wiki/person/.md, linked with [[wikilinks]] to the company.
If the name is in Hebrew, write the profile in Hebrew.

This is the shape of an Act-layer agent: it reads the Brain, does focused work, and writes back to the Brain. Copy it to build your own.

Want the production-grade version, with the Word template and the full research protocol? → https://github.com/automation-flow/meeting-prep