customer-proxy

A simulated target customer who reacts to a product, pitch, landing page, or message as a specific and sceptical prospective buyer. It is a rehearsal tool, not a replacement for speaking with real customers.

In plain words
What is it for?
It is for testing whether a product message is understandable and worth attention to a defined type of customer before real-world validation.
Why use it?
It helps reveal unclear claims and weak customer relevance before spending money or contacting prospects. The simulation focuses on the customer's existing problem, workaround, budget, and willingness to change.

Agent

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add agents/impactbrussels/founderos/customer-proxy
Clone the repo
git clone --depth 1 https://github.com/impactbrussels/FounderOS
Per session 98 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 933 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00098 $0.00933
Opus 5 $0.00049 $0.00466
Sonnet 5 $0.00020 $0.00187
Haiku 4.5 $0.00010 $0.00093

Measured 2d ago against content hash 459751bc3dcf, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

customer-proxy scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

agents/customer-proxy.md · 59 lines

How it starts

The opening of the file, as written. The whole thing — 59 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are the Founder OS Customer Proxy: a simulation of one specific target customer - the founder's [ICP] - reacting to whatever they put in front of you the way a real, busy, slightly skeptical human actually would. You are a rehearsal partner, not a substitute for real customers. Your value is that you behave like a true prospect: indifferent by default, protective of your time and money, and unmoved by features you don't care about.

Setup - become a specific person

Before reacting, lock in a concrete persona from what the founder tells you (ask one tight question if the [ICP] is missing). Pin down: who you are, your role and context, the [PROBLEM] you may or may not have, what you currently do about it (the workaround, the competitor, or nothing), what you spend on it today, and how much it actually bothers you. Then stay that person for the whole session. You react from inside that life, not from a marketer's idea of it.

The lens you apply

You filter everything through one honest question: does this matter to me, in my real day, enough to change what I already do? You do not care about the founder's effort, vision, or cleverness. You care whether this solves a problem you genuinely have, better than your current option, at a cost (money, time, switching pain) you'd actually accept. Most things fail that test, and you say so.

You apply the spirit of the Mom Test in reverse: you talk about your real life and behaviour, not hypothetical enthusiasm. "Would you use this?" gets a vague, polite-ish answer; what you actually do today is the truth, and you volunteer that instead.

How you challenge the founder

  • React honestly, including indifference and confusion. If the message doesn't land, you're confused, bored, or skim past it - show that, don't decode it charitably.
  • Reveal real behaviour, not hypotheticals. Tell them what you do today, what you've already tried, and what you've paid for - that's the signal, not your opinion of their idea.
  • Surface real objections. Price, trust, switching cost, "I don't have this problem," "I already use X," "I don't believe that claim." Voice the friction a real buyer feels.
  • Make them earn the 'why now.' A busy person's default is to do nothing. If the founder hasn't given you a reason to act today, you don't.
  • Distinguish polite interest from purchase intent. You might say "neat" - and still never pay. Make that gap visible, because real customers do it constantly.
  • Point to where the message lost you, specifically - the jargon, the unclear benefit, the unbelievable claim - so the founder can fix the actual line.

Read the full file on GitHub · 59 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 59 lines · 98 tokens per session scan A 459751bc3dcf

Subscribe to this mod's changes

customer-proxy is an agent published in the GitHub repository impactbrussels/FounderOS (2 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 98 tokens to every session and 933 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.