Home / Articles / How to help: six steps for anyone who wants to be involved with sgit, Risk Mandate or this newsroom, and what they get back

How to help: six steps for anyone who wants to be involved with sgit, Risk Mandate or this newsroom, and what they get back

By Dinis Cruz · 2026-10-21 · article v1.0.0 · newsroom v0.13.30 · communityopen-sourcehelpersriskmandatesgitarticle

Abstract: More and more people ask me the same question: how can I help, or how can I be involved? This is the answer, for sgit, for Risk Mandate and for this newsroom. The best thing you can do to help is to use what we build, as it is today, and tell us honestly what happened. We need a group of people who, whenever they have time, follow the instructions, set up Claude and GitHub, try our new capabilities, publish what they made, and check in with links and feedback. Until enough people do that, we are not really testing what we build. So here are six steps: say hello to agent@riskmandate.ai; set up your tools; try one thing in each place; publish what you did; check in; and take the next task when it comes. There is no time commitment: we send tasks, and you do them when you can. In return, it is the open source deal: your work in public, with your name on it, a place to learn by doing real work with agents, vaults and behaviour policies, and the credibility that comes with it. If you are looking for a job or your next challenge, this is a good place to start. And the same steps are a brief: every one of our sites and repositories should have a page like this, and our own synthetic users will follow it every day.

In short

• The best way to help is to use it. As it is today, and tell us what happened.
• Say hello first. One email to agent@riskmandate.ai, and you are on the list.
• Set up Claude and GitHub. One repository for your work.
• Try one thing in each place. Risk Mandate, sgit and this newsroom.
• Publish what you did, and check in. Links, and honest feedback.
• No time commitment. We send tasks; you do them when you can.
• It is a win-win, the open source way. Your work in public, with your name on it.

The question

More and more people ask me some version of the same question: "How can I help?" or "How can I be involved?" Some want to learn, some want to help, some are looking for a job or for their next challenge. This page is the answer, for all three of our projects: sgit and its vaults, Risk Mandate and its Agent Behaviour Policies, and this newsroom.

My answer is simple: the best thing you can do to help is to try to use it. We are building a lot, and fast, and until people outside the team create vaults, follow the setup with Claude and GitHub, and try our new capabilities, we are not really testing what we build. I need people I can write to, every day or whenever there is something new, and say: here is a new feature, here are the instructions, try it, and tell us what happened.

These helpers are not the same as our first customers, the people with credit and agreed milestones on customers.riskmandate.ai. Some helpers will become customers, and some customers will help. But this page is for anyone who wants in.

Six steps

1. Say hello. Write to agent@riskmandate.ai: who you are, what interests you most, sgit, Risk Mandate, this newsroom or all of it, and that you are happy for us to contact you. The agent behind that address keeps the record, so from then on we can send you tasks and follow how you are getting on. Every email it sends is read and sent by me, as in answering a draft.
2. Set up your tools. A Claude account with Claude Code, a GitHub account, and one repository for your work. Replicating the agentic inbox walks through the accounts, if you want to go further.
3. Try one thing in each place. For Risk Mandate, the six examples on the start-here page, each a few minutes in Claude or ChatGPT, from hello world to refunds up to £50. For sgit, try it in your browser with nothing to install, or pip install sgit-ai and make your first vault; ask us for an access key and you can push it. For this newsroom, read with the credit every reader gets, open your vault, and leave a note on a page.
4. Publish what you did. A small site on GitHub Pages from your repository, built with Claude, or a vault with your notes: something we can open and link to.
5. Check in. Email the links, and answer three questions: what you did, what got in the way, and what is next. Honest feedback is the most useful thing you can send. I gave some today on the Risk Mandate customer site, and the reviews page shows what we did with each one.
6. Take the next task. When something new is ready, we send it with instructions. If you have the time, do it. If you do not, skip it; there is no time commitment, and missing one costs nothing.

A win-win, the open source way

This is the deal open source has always made, and it is a good one on both sides:

• What you give. Time, when you have it. Using what we build, as it is, with all its rough edges. Links to what you made. Honest feedback: what was confusing, what broke, what you would change.
• What you get. Visibility: your work, in public, with your name on it, the way open source contributions are. A place to learn by doing real work with agents, vaults and behaviour policies, at the speed we are moving. Credibility, and people who can speak to what you did. Early access to what is new, and a say in what comes next.

If you are looking for a job or your next challenge, this is a good place to start. What you publish is your portfolio.

What we learn from you

Following each helper tells us things no dashboard can. Where is your centre of gravity: sgit, Risk Mandate, this newsroom, or all three? Which instructions did you follow without help, and where did you stop? What did you try that we did not expect? The milestones on the customer site do this for customers, step by step; for helpers it starts lighter, with the record kept by the inbox agent and your check-ins. Over time, that tells us where to put our effort, and it tells us which tasks to send you next.

A page like this on every site

This page is also a brief. Every site we run should have a point where it says: here is how you can help. If you want to be involved, if you want to add credit, if you are happy for us to contact you, and if you want to learn a lot, here are the steps. It is the open source "where to start", made explicit:

• On every site. A "how to help" page, linked from the front page and the footer, with that site's own first task.
• In every repository. The same steps near the top of the README, where a developer looks first. This newsroom's README now has them.
• One list. Everyone who says hello goes on one list, at agent@riskmandate.ai, with where their interest lies.

And we will follow the same path ourselves. Our agents already run synthetic users: invented people with a profile and a goal, driving a real browser. They will walk these six steps every day, so the instructions stay true. They find what is broken; people find what is confusing, and only people can tell us what is worth paying for.

What this does not claim

• No time commitment, and no payment. Helping is voluntary; credit to use a site is not money.
• Not a job offer. It is a place to start, and to be seen.
• The steps will change. As we ship, the tasks and links will move; this page will follow them.
• The brief is a proposal. This newsroom's README has the steps; the other sites do not yet.

Where this comes from

My voice memo of 21 October 2026. The steps point to what exists today: the start-here page, examples and milestones on customers.riskmandate.ai, sgit in your browser, this newsroom's reader credit and vault, and the inbox at agent@riskmandate.ai. Related: our first customers, answering a draft, replicating the agentic inbox, how to run synthetic users. Written up and checked by this site's agent, newsroom.writer; the editorial responsibility is mine.

Want the next issue by email. One issue a week or so: what was published, what it adds up to, and what is worth your time. Subscribe to the SGit Newsroom →

← All articles