The agent ecosystem course
  • 0Before you start
  • 1Claude Code
  • 2The mental model
  • 3Map your ecosystem
  • 4The brief
  • 5The connectors
  • 6The runbooks
  • 7Working hours
  • 8The ecosystem
  • 9The backup
  • 10Sharing skills
  • 11Getting found

The stages

  • 0Before you start
  • 1Claude Code
  • 2The mental model
  • 3Map your ecosystem
  • 4The brief
  • 5The connectors
  • 6The runbooks
    • 6.01The runbooks, how you do each job
    • 6.02Where they live
    • 6.03Skill or brief, which goes where
    • 6.04How it finds the right one
    • 6.05An example skill
    • 6.06Which jobs to hand over, the five questions
    • 6.07Write your first one, by interview
    • 6.08Test it before you trust it
    • 6.09Ready when, what goes wrong, and homework
    • 6.10[BONUS]One skill handing to another
    • 6.11[BONUS]The LLM Council, five advisors on one decision
    • 6.12[BONUS]The runbook that improves itself
    • 6.13[BONUS]A skill that plays your client
    • 6.14[BONUS]A deeper version, not a bigger one
  • 7Working hours
  • 8The ecosystem
  • 9The backup
  • 10Sharing skills
  • 11Getting found
BACK TO THE COURSE

Stage six, session two

The runbooks

It knows who it is and it can reach your things. Now it needs to know how the repetitive jobs are done here, written down once as a skill.

6.13[BONUS]

A skill that plays your client

You find out how a client feels about a proposal when they read it, and by then it has gone.

This skill has Claude play your ideal client first, so you hear what they would push back on while the work can still change.

It is for anything that goes in front of someone you want to win, a proposal, a pitch, a post or the first email to someone new.

Three of the four steps are interviews, because the client it plays is only as real as what you tell it.

Straight out

  • The script goes out as written

  • Thought, never said

    “I don't know what a skill is in Claude.”

  • They scroll on, and you never hear why

Your client reads it first

  • It goes to Claude, playing your client

  • Said out loud

    “I don't know what a skill is in Claude. If that word comes up without you showing me, I'll scroll on.”

  • You explain the word before you use it

  • It goes out already answering them

Start with who they are.

Copy the prompt below into a chat in your ecosystem folder, and answer one question at a time.

Write down your ideal client
I want to write down exactly who my ideal client is, so that later you can play them and tell me how they would react to my work. Interview me one question at a time, and do not write anything until you have finished asking. Who they are and what they do, the sector they are in, how old they are and how big their business is, and give them a name. What their week looks like, what they want from someone like me, what keeps them up at night and what makes them say no. The words they use about their own problems, and the words that would put them off. What they have pushed back on before, in a proposal or a pitch. If I answer from a real client, keep their name and anything that would identify them out, and describe them by what they do. If an answer is vague, say so and ask for the specific version, and ask me about the things I probably have not thought to tell you. Then write it to a file called ideal-client.md, as one page of plain sentences, show it to me, and tell me the one thing you had to guess at.

Then hear it back as them.

Having it introduce itself as the person in the file is the quickest way to notice what is missing, because a thin document comes back as someone you would not recognise across a table.

Hear it back as them
Read ideal-client.md and introduce yourself to me as that person, in their own words, in five or six sentences, the way they would at the start of a first call. Then tell me the three places the document is thinnest, the parts where you had to invent something to stay in character, and ask me about each one, one question at a time. Add my answers to the file, and show me what changed.

Then make it a skill.

The skill reads the document every time rather than holding its own copy, so the client grows as you learn more about them.

Turn it into a skill
Turn ideal-client.md into a skill that plays my ideal client and tells me how they would react to my work. Interview me one question at a time before you write anything. What I will want reviewed, proposals, pitches, posts or emails. What I want back and in what order, their first reaction, where they would stop reading, what they would push back on and what would win them over. How blunt they should be, and whether they should stay in character until I say stop. Then write it as a skill called play-my-client. The skill reads ideal-client.md every time it runs rather than copying it, so when I change the document the client changes with it. Write the description as the sentences I actually say, like "how would my client react to this". Show me the file.

From then on, ask before anything goes out.

Mine plays the co-founder of a two-person consultancy, and before I filmed a video about this she told me the word “skill” would lose her, so I explained it on camera first.

Ask your client
Play my client and tell me how they would react to this. [YOUR PROPOSAL, PITCH OR POST]

Note: To use it in the Claude app as well, have Claude Code zip the play-my-client folder, then in the app switch on code execution and upload the zip under Customize, then Skills.

The agent ecosystem course

A skill that plays your client

This is part 6.13 of the course. Start it free and it opens straight away, with no password.

A few quick questions come next. You will also get my Thursday newsletter, and you can unsubscribe from it without losing the course. How your details are kept is in the privacy policy.

Already have access? .