The short version: we collect what running a metered model platform requires — who you are, what you spent, and what you sent — and nothing beyond it. Your prompts are not training data and are not for sale.
1. Who this covers
This page describes how Agent Platform, reachable at agentplatform.store, handles personal data when you use the model gateway, a hosted runtime, or a workspace.
It applies to the account holder and to anyone acting under that account. Where a customer of yours interacts with an agent you deployed, you are the one deciding what happens to their data; we process it on your instruction.
2. What we collect
Account identity: the address or wallet you sign in with, and whatever display name you set.
Billing records: top-ups, deductions and the balance that results, together with the on-chain references where a transaction settled on chain.
Usage metering: for each request, the model, the token counts, the route taken, the latency, and the resulting charge.
Content you submit: prompts, completions, files and configuration you place in a runtime or workspace.
Operational telemetry: error traces and coarse request logs that let us tell a broken route from a broken prompt.
3. What we use it for
Running the service you asked for, metering it accurately, preventing abuse, and answering you when something goes wrong.
Aggregate, non-identifying figures inform capacity planning and which routes we keep. Nothing in that aggregate is traceable back to a prompt.
4. Model processing
A request is held in memory long enough to route, execute and meter it, then released. We do not retain prompt or completion bodies for training, and we do not sell them.
The upstream vendor that actually serves a model receives the request and handles it under their own terms. Where a key uses private routing, the payload passes through an attested enclave that we cannot read, but the vendor on the far end still receives it.
5. Who else sees it
Model vendors, for the calls you direct to them. Infrastructure providers that host the runtimes and workspaces. Payment and chain infrastructure, for settlement.
We do not sell personal data, and we do not share it with advertisers.
6. How long we keep it
Metering and billing records are retained for as long as the account exists plus the period our accounting obligations require.
Content inside a runtime or workspace lives until you delete it or delete the resource holding it. Deleting a workspace destroys its volume.
Operational logs roll off on a short cycle.
7. Security
Each runtime is an isolated container with its own encrypted volume. Secrets are encrypted at rest and injected at process start rather than written into configuration files.
Access to production is limited and audited. No control is absolute, and we will tell you plainly if something affecting your data goes wrong.
8. Cookies and local storage
The site stores your theme choice and, if you connect one, the identifier of the wallet you used, so a returning visit does not prompt again. Both live in your browser and are readable only by this site.
There is no advertising or cross-site tracking on this domain.
9. Your choices
You can export your usage records, delete resources you created, and close the account. Depending on where you live you may also have rights to access, correct, or object to certain processing.
Write to the contact address below and we will act on it rather than route you through a form.
10. Children
The platform is not intended for anyone under sixteen, and we do not knowingly hold data about them.
11. Where data goes
Runtimes and workspaces run in the region you pick. Model calls may be served from a different region, because that is where the vendor's capacity sits.
12. Changes
When this page changes materially we will say so on the site before the change takes effect, and the date at the top will move.
13. Contact
Reach us through https://x.com/agentplatform_, or at the address published on agentplatform.store.