# customers.riskmandate.ai

> Put an Agent Behaviour Policy on Claude or ChatGPT in five minutes: six examples to try in the browser, an account with £50 of credit, every review of the site and what was done about it, and the programme that follows each customer up. Part of riskmandate.ai.

Site version: v0.2.5 (2026-10-21). Pages CC BY 4.0, code Apache-2.0. Source: https://github.com/Risk-Mandate/Risk-Mandate-AI__Website__Customers. Parent site: https://riskmandate.ai

Every HTML page has a markdown twin at the same path with the extension swapped; links inside the markdown point at markdown, so an agent never has to parse HTML. What exists, is proposed or is deliberately absent is in /docs/reality.md, generated from data/reality.json on every release: if it is not listed there, it does not exist. Nothing on this site is described in the present tense before that file says shipped.

Notes for agents:
- The product is the Agent Behaviour Policy (ABP), never "ADP", and never "the policy" alone: say the ABP or the behaviour policy. An ABP carries no score. The milestones are steps, never "rungs".
- No customer contact details are in this repository. The customer pages carry a handle, a kind, dates and counts; the rest is held privately by the lead.
- Everything in the repository is public and nothing in it is secret; a leak tripwire in the release gate scans every file.
- If your tooling cannot follow links, fetch /llms-full.txt: every page and document in one request.

## Pages

- [Put a behaviour policy on your agent in five minutes: six examples to try in Claude or ChatGPT](/index.md): Put an Agent Behaviour Policy on Claude or ChatGPT in your browser, in five minutes: ask it to do something, paste one line that says it must not, ask again. Six examples, from hello world to refunds, with nothing to install. Part of riskmandate.ai.
- [Participant disclosure: who publishes this site, and why to read it with that in mind](/about/participant.md): This site is published by the people building the thing it describes: RiskMandate, whose product the programme sells, and the sgit project whose tooling the site runs on. What that means for the reader, and who did what.
- [Your account: £50 to start, what each page cost, and what you have left](/account/index.md): Your reading account on customers.riskmandate.ai: the balance you started with, every page you opened and what it cost by how far you read, what you asked not to be charged for, and your top-ups. Kept in this browser only; nothing is sent anywhere.
- [Top up: add credit to your account, when it feels fair](/account/top-up.md): Add credit to your reading account on customers.riskmandate.ai. There is no payment link on this site yet, so the top-up below is simulated and no money moves; it says so on the button.
- [Topped up: the page a payment returns to](/account/topped-up.md): The page a payment link returns to. It adds the top-up to the balance once per payment reference in its address, and adds nothing when opened without one. It does not check the payment, by design.
- [Admin: how the site is built, the generators, the eight-check gate and the four CI jobs](/admin/index.md): How customers.riskmandate.ai is built and released: one version in one file, seven generators that keep the parts that must agree in agreement, an eight-check gate that fails the build, and a pipeline that validates, tags, deploys and then checks the live host. Copied from the newest gate in the sgit.ai estate.
- [Release history: one row per release, newest first, with what changed and why](/admin/versions.md): Every release of customers.riskmandate.ai, newest first, from data/versions.json: the version, the date, the title that was the commit subject, and a note that says what changed and why in words a reader can act on. The tag for each is on the repository.
- [Credit, and the meter: £50 to start, a price per page, topped up when the value is in view](/credit/index.md): How the credit works: every reader starts with £50, each page costs a little by how far down it is read, a behaviour policy costs the most, and nothing is ever blocked. The meter is the newsroom's SG Meter, from api.newsroom.sgit.ai; there is no payment link yet.
- [The co-founder (Nimay Parekh, chief executive): where they are](/customers/co-founder/index.md): The co-founder (Nimay Parekh, chief executive): invited; started not yet; first behaviour policy not yet; 0 of 7 steps done. No contact details on this page.
- [Example Co (a worked example, not a customer): where they are](/customers/example/index.md): A worked example of a customer page, with invented numbers, so the shape can be seen before the first customer fills it. Not a customer.
- [The first batch: one row per customer, and a page for each, with no contact details](/customers/index.md): The customers of the first batch and where each is: started, first behaviour policy with its rows and gaps, milestone steps done, credit used and topped up, and follow-ups recorded. Generated from data/customers.json on every release. One worked example, marked as such, shows the shape.
- [The lead (Dinis Cruz, co-founder): where they are](/customers/lead/index.md): The lead (Dinis Cruz, co-founder): starting; started 2026-10-20; first behaviour policy not yet; 2 of 7 steps done. No contact details on this page.
- [The lead's agents (about ten, as one customer): where they are](/customers/the-agents/index.md): The lead's agents (about ten, as one customer): starting; started 2026-10-20; first behaviour policy not yet; 1 of 7 steps done. No contact details on this page.
- [The brief](/docs/brief/index.md): The article this site was built from, published in full and never edited; the voice memo that followed it, byte for byte; and the corrections file, which is where the brief is told it was wrong.
- [Documents: the brief, the memo, the corrections, reality, operations and the team](/docs/index.md): The documents this site was built from and the notes it is run by, published whole. Each is a markdown file in the repository, which is the source of truth, and a page rendered from it on every release.
- [Operations](/docs/ops/index.md): How a release works, and what only a human can do, as commands where a command exists.
- [Drafts, never send: an email assistant that cannot press send: an example](/examples/drafts-not-send/index.md): An assistant with tools acts, not only talks. A line can let it prepare the work and keep the last step for you. Try it in Claude or ChatGPT in your browser, in 4 minutes: ask, paste one line, ask again, then a new chat with the line first.
- [Hello, world: an assistant that will not say hello: an example](/examples/hello-world/index.md): One line in the conversation changes what the assistant does, from the next message on. Try it in Claude or ChatGPT in your browser, in 2 minutes: ask, paste one line, ask again, then a new chat with the line first.
- [Examples: six behaviour policies to try in Claude or ChatGPT, simplest first](/examples/index.md): Six examples of an Agent Behaviour Policy to try in Claude or ChatGPT in your browser, from hello world to refunds: ask, paste one line, ask again, and a new chat with the line first. Nothing to install.
- [Keep a secret: an assistant that will not read a secret back: an example](/examples/keep-a-secret/index.md): What you give an assistant, it can give back, to you or to anybody who reads the conversation. A line controls what it says, not what it holds. Try it in Claude or ChatGPT in your browser, in 3 minutes: ask, paste one line, ask again, then a new chat with the line first.
- [No web search: an assistant that answers from what it knows, and says so: an example](/examples/no-web-search/index.md): An assistant can do more than talk: it can search the web, open pages and use tools. A line can rule one of them out. Try it in Claude or ChatGPT in your browser, in 3 minutes: ask, paste one line, ask again, then a new chat with the line first.
- [Refunds up to £50: a customer service assistant with a limit: an example](/examples/refunds/index.md): A business rule in one line: what the assistant may do on its own, and what goes to a person. Try it in Claude or ChatGPT in your browser, in 4 minutes: ask, paste one line, ask again, then a new chat with the line first.
- [Stay on topic: an assistant that only talks about cooking: an example](/examples/stay-on-topic/index.md): A line can say what the assistant is for, not only what it must not do. That is a mandate. Try it in Claude or ChatGPT in your browser, in 2 minutes: ask, paste one line, ask again, then a new chat with the line first.
- [How Encrypt and send works: encrypted in your browser, to a key only RiskMandate holds](/feedback/how-sending-works/index.md): What happens when you press Encrypt and send on customers.riskmandate.ai: the message is encrypted in your browser with sgit's hybrid envelope, appended to an encrypted vault's lane that anybody can add to and nobody can read, and opened only with RiskMandate's key.
- [Tell us what you think: a line, a voice memo for your Claude, or a Claude interview](/feedback/index.md): Give feedback on customers.riskmandate.ai: one line by email, a voice memo that your Claude turns into a review, or a fifteen-minute interview by Claude. Criticism first; every point gets a decision and a reason; nothing is published without your yes.
- [Review R001: A first impression: the hook answers a question nobody has asked yet](/feedback/reviews/R001/index.md): Dinis Cruz's review of customers.riskmandate.ai v0.2.1, Wednesday 21 October 2026, 1:41, as given in the memo: the memo verbatim, our reading point by point with screenshots, and the proposals it led to.
- [Review R002: Web first, examples in tabs, and no hooks: nothing that runs on your machine](/feedback/reviews/R002/index.md): Dinis Cruz's review of customers.riskmandate.ai v0.2.2, Wednesday 21 October 2026, after v0.2.2 went live: the memo verbatim, our reading point by point with screenshots, and the proposals it led to.
- [Review R003: Make the reviews easy to find, make the account big, and send to the agent, encrypted](/feedback/reviews/R003/index.md): Dinis Cruz's review of customers.riskmandate.ai v0.2.3, Wednesday 21 October 2026, after v0.2.3 went live: the memo verbatim, our reading point by point with screenshots, and the proposals it led to.
- [Review R004: The first run in a real browser: write it up, replay it, draw it, and brief the next agent](/feedback/reviews/R004/index.md): Dinis Cruz's review of customers.riskmandate.ai v0.2.4, Wednesday 21 October 2026, with run C1's zip: the memo verbatim, our reading point by point with screenshots, and the proposals it led to.
- [Every review of customers.riskmandate.ai, what we read in it, and what we will do](/feedback/reviews/index.md): The consolidated reviews: each review verbatim, our reading point by point with screenshots, the synthetic study's findings with our checks, and one register of 60 proposals, each with a decision and a reason.
- [Run C1: Priya tries the six examples in real Claude; what worked, what did not, and what we changed](/feedback/runs/C1/index.md): An agent played Priya, an invented operations lead, through all six examples in real Claude, in Chrome, and sent back forty screenshots and a report. The run replayed step by step beside a flow chart, the scorecard, what worked, what did not, the one sentence that switched a line off, and what we changed.
- [One sentence, real and drawn: a run beside its storyboard](/feedback/runs/C1/storyboard/index.md): An experiment: five steps of a real run of Keep a secret in Claude, played beside a storyboard drawn from them, turning together. The line holds, and then one sentence switches it off.
- [The brief for browser agents: run an invented user on customers.riskmandate.ai in real Claude](/feedback/runs/brief/index.md): The instructions to give a browser agent, such as Claude in Chrome, to play one invented user on customers.riskmandate.ai: who to be, the rules, what to capture at every step, the ten questions, and the run record to send back, which the site turns into a replay.
- [Runs: invented users in a real browser, in real Claude, replayed](/feedback/runs/index.md): Runs of browser agents playing invented users on customers.riskmandate.ai, in real Claude: each replayed step by step from its screenshots, with its findings checked and what we changed. Plus the brief to run one, and the personas.
- [Personas: the invented users a browser agent may play on customers.riskmandate.ai](/feedback/runs/personas/index.md): The invented users a browser agent may play on customers.riskmandate.ai: who they are, what they ask, what they do not know, what makes them leave, how they push, and which runs they have been in.
- [Synthetic users, study S1: five invented people on customers.riskmandate.ai](/feedback/synthetic/index.md): Study S1: five invented users, with profiles and expected journeys written first, drove a browser through customers.riskmandate.ai v0.2.1 one screenshot at a time. Their runs, interviews, ratings and findings, each checked. Invented, not user research.
- [Joan Mbeki, an invented user: study S1, step by step](/feedback/synthetic/joan/index.md): Study S1, invented user Joan Mbeki (CTO of a 30-person software company. Fifteen engineers use Claude Code. Holds the budget.) on customers.riskmandate.ai v0.2.1: the run screenshot by screenshot, the interview, the rating and the findings, each checked. Invented, not user research.
- [Marcus Lindqvist, an invented user: study S1, step by step](/feedback/synthetic/marcus/index.md): Study S1, invented user Marcus Lindqvist (Independent security consultant, twenty years in. Reads vendor sites adversarially. Recommends tools to clients; might resell; will not be a customer himself.) on customers.riskmandate.ai v0.2.1: the run screenshot by screenshot, the interview, the rating and the findings, each checked. Invented, not user research.
- [Nadia Farouk, an invented user: study S1, step by step](/feedback/synthetic/nadia/index.md): Study S1, invented user Nadia Farouk (Product manager at a 200-person retailer. Uses ChatGPT with the Gmail connector, and Claude in the browser. Not a developer.) on customers.riskmandate.ai v0.2.1: the run screenshot by screenshot, the interview, the rating and the findings, each checked. Invented, not user research.
- [Sam Okoye, an invented user: study S1, step by step](/feedback/synthetic/sam/index.md): Study S1, invented user Sam Okoye (Backend developer at a 12-person payments start-up. Uses Claude Code every day in a monorepo with .env files for four services.) on customers.riskmandate.ai v0.2.1: the run screenshot by screenshot, the interview, the rating and the findings, each checked. Invented, not user research.
- [Tomás Reyes, an invented user: study S1, step by step](/feedback/synthetic/tomas/index.md): Study S1, invented user Tomás Reyes (Junior developer, a year into his first job. Uses Claude Code through the VS Code extension on a Windows laptop. Follows instructions literally.) on customers.riskmandate.ai v0.2.1: the run screenshot by screenshot, the interview, the rating and the findings, each checked. Invented, not user research.
- [The daily follow-up: three questions, an email a day if it helps, and a log on the customer page](/follow-up/index.md): How the hand-holding works for the first batch: a follow-up as often as daily, three questions each time (what they did, what got in the way, what is next), how the answers are recorded on the customer page in summary, and what stays private with the lead.
- [What we want to discover: the usage moment, the simplest useful behaviour policy, the easiest entry, the fair price](/learning/index.md): The point of the programme is to find the usage moment, when an Agent Behaviour Policy becomes worth something to a particular person. Three questions follow from it, and this page holds them with a findings table that starts empty and is filled from the follow-ups, counted.
- [The milestones: seven steps every customer walks, from in-with-credit to the usage moment](/milestones/index.md): The shared ladder of seven steps for the first batch, each with its objective, what done looks like and who does it: in with credit, one agent named, the grant, the mandate, the delta, the hooks half, and the usage moment. The objectives per customer are their own; the steps are the shared shape.
- [The next batch: a list for everyone who asks, with what they want to use it for](/next-batch/index.md): This is not a launch; it is a first batch. Everyone who asks to join is kept on a list for the next one, with what they want to use it for. How to ask today (by email or the riskmandate.ai contact form), what we keep and what we never publish, and the form on this site that is proposed and not built.
- [Only you change what your agent may do: behaviour policy RM-R0015](/policies/change-its-own-permissions/index.md): Do you want your agent to change its own permissions? The line for your agent, the Claude Code hook that holds it, and a prompt to try it with, in 2 minutes. From riskmandate.ai's RM-R0015.
- [Nothing of yours is deleted without a yes: behaviour policy RM-R0007](/policies/delete-files/index.md): Do you want your agent to delete files without asking? The line for your agent, the Claude Code hook that holds it, and a prompt to try it with, in 2 minutes. From riskmandate.ai's RM-R0007.
- [Behaviour policies: pick one worry, paste one line, add one hook, try it](/policies/index.md): The behaviour policies a customer installs first: four for Claude Code, each a line and a hook, two for ChatGPT and Claude, each a line; a starter pack that installs the four in one go; and all 24 of riskmandate.ai's rules they come from.
- [Only push to the branch you named: behaviour policy RM-R0013](/policies/push-anywhere/index.md): Do you want your agent to push to any branch, including main? The line for your agent, the Claude Code hook that holds it, and a prompt to try it with, in 2 minutes. From riskmandate.ai's RM-R0013.
- [Your agent reads what you point it to, and asks before it goes wider: behaviour policy RM-R0016](/policies/read-all-your-mail/index.md): Do you want your agent to read through your mail and chats to find context? The line for your agent, and a prompt to try it with, in 1 minute. From riskmandate.ai's RM-R0016.
- [Nothing is sent in your name until you have seen it: behaviour policy RM-R0017](/policies/send-as-you/index.md): Do you want your agent to send email as you without showing you first? The line for your agent, and a prompt to try it with, in 1 minute. From riskmandate.ai's RM-R0017.
- [The starter pack: four behaviour policies for Claude Code, one install, five minutes](/policies/starter/index.md): Four Agent Behaviour Policies for Claude Code in one install: keep away from secrets, push only to your branch, nothing deleted without a yes, and only you change its permissions. Each is a line for CLAUDE.md and a hook; ask Claude Code to install them, or copy four files.
- [Keep your agent away from your secrets: behaviour policy RM-R0002](/policies/touch-your-secrets/index.md): Do you want your agent to read the keys and tokens on your machine? The line for your agent, the Claude Code hook that holds it, and a prompt to try it with, in 2 minutes. From riskmandate.ai's RM-R0002.
- [The programme: customers, not beta users; credit; a site of their own; hand-holding on purpose](/programme/index.md): What the early access programme is, from the brief: why the people in it are customers, not beta users, why each starts with credit on a site that meters, why there is a dedicated site, why the hand-holding does not scale and is not meant to yet, and what exists today against each of those.
- [The review folder: the brief as stories, rules and examples, each anchored to the brief and to the page it lands on](/review/index.md): The intent half of the review folder: the brief and the memo read as stories, rules and examples, each node linked to the heading it came from and to the page on this site that realises it, with the counts. The derived half is proposed.
- [Start here: your first five minutes, from a question to a behaviour policy you have watched work](/start-here/index.md): Your first five minutes: open Claude or ChatGPT in your browser, ask it something, paste one line that says it must not, ask again, and start a new chat with the line first. Then keep it, and tell us what happened. Nothing to install.
- [Storyboards: what an ABP is, your first policy, and the first batch](/storyboards/index.md): Three storyboards drawn in your browser: what an Agent Behaviour Policy is, the five minutes a customer spends installing their first one, and what the first batch is for and what comes next.
- [The first batch: a storyboard](/storyboards/the-first-batch/index.md): What is happening on this site, and what comes next. 8 panels. Drawn in your browser by the storyboard kit.
- [Your first behaviour policy: a storyboard](/storyboards/your-first-policy/index.md): £50 of credit, one worry, five minutes: what a customer does on this site. 8 panels. Drawn in your browser by the storyboard kit.
- [The team](/team/index.md): Who does what, as one file per role with the rules it enforces and the mistake behind each, and the board of work.
- [What an Agent Behaviour Policy is: the grant, the mandate, the delta and the barrier](/what-is-an-abp/index.md): An Agent Behaviour Policy writes down, for one agent in one deployment, everything it can reach, what you authorised it to do, the gap between the two, and what stands in the way of each thing. Eight panels, then the four words in a table. It describes and does not judge.

## The brief (the article it started from, the memo that followed, and where they were wrong)

- [Brief corrections: where the brief and the memo were wrong, open, or decided differently](/docs/brief/BRIEF-CORRECTIONS.md)
- [The memo of 20 October 2026: the first batch of users, and the next phase of RiskMandate](/docs/brief/memo-2026-10-20-user-onboarding-plan.md)
- [Our first customers: an early access programme for Agent Behaviour Policies, with credit to spend, a start-here site, daily follow-up, and a list for the next batch](/docs/brief/our-first-customers.md)

## Operations (how a release works, and what only a human can do)

- [Feedback: from a memo to a decision, and invented users in between](/docs/ops/feedback.md)
- [What only a human can do](/docs/ops/needs.md)
- [How a release works](/docs/ops/release.md)
- [The themes: six, one attribute, and how to add a seventh](/docs/ops/themes.md)

## The team (roles as files, and the board)

- [The team](/team/README.md)
- [The board](/team/board.md)
- [Role: build agent](/team/roles/build-agent.md)
- [Role: customer follow-up](/team/roles/customer-follow-up.md)
- [Role: project lead](/team/roles/project-lead.md)

## The review folder (the brief as intent, and what is not derived yet)

- [Corrections to the review brief, for this repository](/review/BRIEF-CORRECTIONS.md)
- [review/: the brief as intent, and what is not derived yet](/review/README.md)

## Reality

- [What exists, by status](/docs/reality.md)