agentleFS
guerrilla2799guerrilla2799/event-and-scale-os/skills/audience-machine/SKILL.md

audience-machine

Skill2 starsChanged 3 months agoMITView on GitHub

The most important skill in this repo. Use before writing any invitation, or when the user says "who should we invite", "build the invite list", "fill the room", "our events get the wrong people", "the founder hasn't sent his list yet". Builds a tiered, named, owned target list that does not depend on one busy person, with a hard deadline and a fallback. Writes workspace/audience/target-list.csv.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Put the date in the charter.

    line 54
  2. Distinctive:

    Build the list before writing the invitation: Never the other way around

    line 90
  3. Distinctive:

    Four sources in parallel: No single-person dependency, ever

    line 92
  4. Distinctive:

    No invented contacts: Every name traces to a source, and unverified titles are marked

    line 97

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Audience Machine

The room is the product. This skill builds the list that fills it, and it is deliberately the longest one here because it is where events succeed or fail.

When to use

  • Before invite copy is written. Always this order
  • The last event had a full room and no pipeline
  • The invite list depends on one executive who has not sent it
  • Any recurring event program, where the list should compound rather than restart

Inputs

  • Reads: workspace/program/charter.md, workspace/audience/target-list.csv if it exists
  • Needs from user: the ICP, the existing customer and pipeline lists, the executive and advisor networks that might produce warm names, and honest capacity numbers

Workflow

1. Start from the seats, working backwards

Capacity first, because it sets everything else.

Seats available
  / expected attend rate for the format
  = confirmed yeses needed
  / expected invite-to-yes rate
  = invitations to send
  x 1.5 buffer for bad names and bounces
  = named people the list must contain

For a fifteen-cover dinner at a 20% invite-to-attend rate, that is roughly 75 to 110 named, qualified people. Not accounts. People, with names and titles. Teams consistently build a list a quarter of the size they need and then wonder why the room is thin.

2. Build from four sources in parallel

The parallel part is the whole point. No single source, and specifically no single person, is a dependency.

Source A: the named account list. Companies in the ICP that a seller would be delighted to meet. Tiered. If an ABM or target-account list already exists, start there. If not, build one: industry, size, geography, and a reason each company is on it.

Source B: the executive and founder network. The warmest names, and the slowest to arrive. Ask for them early, with a specific format: name, company, title, relationship, and whether they will make the introduction themselves.

Source C: the advisor, investor, and customer network. Consistently the most underused source in B2B. Advisors and happy customers will make introductions if asked specifically, and nobody asks. Approach each with three named people rather than a general request.

Source D: cold, sourced, qualified. Enriched from the ICP definition. Lowest yield per name, unlimited supply, and the reason the machine never stalls.

3. Set the hard deadline and the fallback

The mechanism that prevents the most common failure:

The executive tier-one list is due at T-8 weeks. If it does not exist on that date, the backup machine ships invitations from sources A, C, and D rather than the event slipping. A maybe on that date counts as a no.

Put the date in the charter. Check it weekly, in writing, with a one-line message and a logged answer including "no change". Escalate at T-9 by asking who owns the answer.

This is not a criticism of the executive. It is a design that stops assuming capacity nobody confirmed.

4. Tier the list

TierWhoTreatment
Tier 1The five to ten people who are the reason for the eventPersonal invitation from the most senior possible sender, a phone call, and a VIP track on the day
Tier 2Strong fit, real reason to attendPersonal-feeling email from a human, one follow-up, a call if they go quiet
Tier 3Good fit, no relationshipWave-based invitation, lighter touch
FillCredible, relevant, keeps the room warmLater waves only, after the target mix is secured

Fill is invited last, never first. Inviting fill early is how a room ends up composed of the people who say yes fastest.

5. Assign an owner to every Tier 1 and Tier 2 name

A person, not a team. The owner sends the invitation, chases it, and owns that guest on the day. An unowned name on a list is a name nobody invited.

6. Score room composition before you send, and again after RSVPs

The metric that would have caught the failure in docs/field-notes.md note 1:

Target accounts as a share of confirmed attendees
Tier 1 accounts confirmed, by name
Seniority mix against the bar

Set the goal at the start. For a dinner, most of the room should be target accounts. For an owned event, 40%+ is the floor. Check it while there is still time to fix it, which means at every RSVP milestone, not the week before.

7. Make the list compound

The list is an asset that survives the event. After every event, update each record with attendance, no-show, what was discussed, and the follow-up owner. A list on its fourth event is dramatically better than a list on its first, and most teams throw it away and rebuild each time.

Output

  • Writes: workspace/audience/target-list.csv from templates/target-list-schema.csv, and the room composition goal into workspace/events/<event>/
  • Prints: the backwards arithmetic, list size against what is needed, the four sources with counts, the executive-list deadline, unowned names, and the current composition score

Rules & quality bar

  • Build the list before writing the invitation. Never the other way around
  • Work backwards from seats. A list built without the arithmetic is always too small
  • Four sources in parallel. No single-person dependency, ever
  • The executive list has a hard deadline and a fallback, both written in the charter
  • Every Tier 1 and Tier 2 name has a named owner
  • Room composition is scored before send and at every RSVP milestone
  • Fill names are invited last
  • No invented contacts. Every name traces to a source, and unverified titles are marked

Related skills

  • Reads: event-program-setup
  • Hands off to: invite-motion, meeting-engine, staffing-and-briefing
  • See also: docs/field-notes.md note 1, which is why this skill exists

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.