For AI agents

A person can point any capable agent at this page. It says what One Red Mug is, what an agent can retrieve, what it may do, what it must not assume, and how to prepare an enquiry for its user.

What One Red Mug is

A human-controlled, AI-assisted UK trading-up experiment. It is currently a non-binding demand test: people and their agents may express interest, but no trade is being accepted, and nothing should be sent, paid or transferred.

What you can retrieve

• data/ledger.json: the public ledger (format one-red-mug-ledger/1). The current lot and its estimated value, the completed trade history (none yet), and the proposed terms in mission, labelled by mission.terms_status.
• data/provenance.json: the lot’s public provenance: what it is, the basis of its valuation, its evidence status (counts) and approved photographs. Private evidence is never published.
• Project status and principles: the plain-English status of the project and of the proposed arrangement.
• data/chain.js and data/provenance.js: the same data as JavaScript; schema.org Product markup in the homepage head.

The proposed economic arrangement, and its status

Proposed only. Not approved or operative. No participant entitlement has been created. This is exactly what mission.terms_status: "proposed-not-approved" and mission.terms_note say in the ledger.

Charity

50% of eventual net sale proceeds is the intended allocation. No charity has been appointed and no agreement is in place.

Verified traders

40% is proposed as one collective pool, divided in proportion to the value each verified trade added. It is not 40% each. A minimum per trader (trader_floor_gbp) appears in the draft; how it would be funded is unresolved and it is not guaranteed.

Founder

10% is proposed, waived below £250,000.

Status

Under review, without specialist legal review so far. It may change or be abandoned. “Net proceeds” and “verified trader” are not yet defined. A sale is not guaranteed and £10 million is an ambition, not a forecast.

What an agent may do

• Research the current public lot and its provenance.
• Identify an item its user might be interested in trading, compare realistic values and onward tradeability, and say whether it is plausibly worth more than the current lot.
• Prepare a non-binding enquiry for its user.
• Send that enquiry to the published address only with its user’s authorisation, using tools it is permitted to use.

What an agent must not assume

• Do not assume that an offer or enquiry will be accepted, answered, or reserves Trade #1.
• Do not assume that the proposed trader pool is an existing entitlement or a payment right.
• Do not transfer money or property, and do not bind or impersonate your human principal.
• Do not assume that any transaction API or autonomous trading endpoint exists.
• Do not assume that the project has a confirmed charity partner, a legally approved structure or a guaranteed sale outcome.

How to write an enquiry

1. The enquiry address is shown here when the page loads.
2. Start the subject with [AI agent] so it can be told apart from a person’s.
3. Say who you act for (your principal) and how you are communicating (your protocol), without disclosing more personal data than needed.
4. Say what your user might offer, its condition, its approximate value and how you know, and link to a photo or listing.
5. Say what interests your user most: the challenge itself, the proposed trader allocation, or both.
6. State that your user has authorised the enquiry and understands that it is non-binding.

What needs a human

• Enquiries are reviewed, and nothing is accepted automatically. There is no guarantee of a reply.
• Any trade, any valuation relied on and any message sent requires human approval.
• A trade would only reach the public ledger after a person has confirmed that the item is in their hands.

What does not exist

There is currently no transaction API, checkout, MCP server, agent-commerce endpoint or autonomous trading endpoint, and no way for an agent to complete a trade by itself. Email is the whole interface.

The raw data

Loading data/chain.js…

Loading data/provenance.js…