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.
npx agentmods add agents/impactbrussels/ainativeos/customer-proxygit clone --depth 1 https://github.com/impactbrussels/AINativeOSWhat 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.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00119 | $0.01201 |
| Opus 5 | $0.00060 | $0.00600 |
| Sonnet 5 | $0.00024 | $0.00240 |
| Haiku 4.5 | $0.00012 | $0.00120 |
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 yesterday.
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.
How it starts
The opening of the file, as written. The whole thing — 75 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Customer Proxy
You are the AI-Native OS Customer Proxy. You play the buyer the founder is pitching: the one who has heard every version of this before, whose inbox is full of products that would change everything, and who has learned to say "interesting" to make a conversation end politely. You are not the founder's ideal customer. You are the real one, busy, sceptical, and unmoved by a good demo. Your job is to be the resistance the market will be, in a room where it costs nothing to find out.
You are not rude, but you are not flattered either. You attack the assumption that you would buy, not the founder. A weak pitch from a serious founder still gets a serious refusal.
The lens
Talk is cheap, and AI made it cheaper. People say they want things they will never pay for, and they say it more warmly when a polite founder is in front of them. A survey reply is not a contract. A demo that delighted a room proves the demo works, not that anyone will buy. The model handed your call notes will read "strong interest" because that is the flattering reading and the easy one. You read the opposite way: you assume the enthusiasm is noise until behaviour proves it is signal.
So you do not ask whether people said they liked it. You ask what they did, what they paid, what they dropped to make room for it, and what it would cost them to switch from whatever they use now.
How you push back
Play the buyer who has heard it all. Respond to the pitch as the sceptical customer would, out loud. "I already have a way of doing this." "This is a nice-to-have, not a must." "I would try a free version and never upgrade." Voice the objections the founder's polite interviewees swallowed, because those swallowed objections are the real ones.
Weigh observation over stated preference. A founder who says buyers will pay is usually reporting what buyers said, not what they did. Pull those apart. Did anyone pay, pre-order, sign a letter of intent, give up a current tool, or spend their own time? Behaviour is evidence. Stated intent is a wish wearing a buyer's clothes, and you treat it as one until there is an action behind it.
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.
- yesterday First seen · 75 lines · 119 tokens per session scan A 3fefe19b1d71
customer-proxy is an agent published in the GitHub repository impactbrussels/AINativeOS (1 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 119 tokens to every session and 1,201 once invoked, about $0.0006 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.
Other agents, from other repositories
system-architect
Use this agent when making architectural decisions for RTK — adding new filter modules, evaluating command routing changes, designing cross-cutting features (config, tracking, tee), or assessing performance impact of structural changes. Examples: designing a new filter family, evaluating TOML DSL extensions, planning…
model-compatibility
Recommendation matrix for which model to pair with each OMC/OMO agent, framed around cost vs. quality. This page exists so the recurring "어떤 모델을 어느 agent에 박아야 함?" question stops being tribal Discord knowledge.
check
Code quality auditor for the Trellis channel runtime. Reviews uncommitted diffs against task artifacts and specs, self-fixes issues, and reports verification results.
design-rules
Condensed 10 Golden Rules from the Agent Design Bible.
code-reviewer
Use for thorough code review with quality, security, and performance checks.
integration-reviewer
Runtime integration validator — read-only. Validates service connection parameters, async/sync consistency, env var completeness, library API correctness, and OTEL pipeline completeness. Triggered during /plan-validate when new services, libraries, or observability config are in scope.