Quick answers to rent or publish your first agent.
A marketplace of specialized AI agents. Each agent is built by a creator who knows the subject (employment, sales, paperwork, finance…). You activate your own copy and use it by chat.
No. Pick an agent, activate a plan and write to it like you would to a person.
We do not use passwords. Type your email in “Sign in” and we send you a link that lasts 15 minutes and works once. If it does not arrive, check spam or request another: you can request up to 5 per hour.
Each agent shows its plans on its page, in Chilean pesos. Many have a free plan with a daily message limit. Paid plans turn on when we enable payments.
The free plan limit resets at midnight (Chile time). Once payments are enabled you will be able to switch to the agent’s paid plan from the chat.
No. The creator does not see your data or your credentials. For agents with the “External” badge, Agnose forwards your messages with a pseudonymous id, never your email or name.
The agent runs on its creator’s infrastructure and Agnose is the gateway (limits, history and billing). If that server fails several times in a row, we pause the agent.
Yes. While we add the button to “My agents”, write to support and we will cancel it for you.
Use “Report a problem” in the chat or write to support. Every agent version is reviewed before it is published.
Sign in, create your profile in “Create” and then your agent. The version can come from a zip with agnose.json, a GitHub release or your own endpoint (own server or A2A). Then send it to review.
Security, the permissions it asks for, and that what the page promises is true. If we return it, you will see the note on the agent’s page.
80 % of what renters pay for your agent, or 85 % if it runs on your servers. It applies once payments are enabled.
No. They only use the agent by chat; they have no access to the machine where it runs.
Nothing changes for what is published: the snapshot of that commit is what was approved. A change is a new version that goes through review.
It failed the health check 3 times in a row. Check your endpoint: it reactivates by itself once it responds again.
Write to us and we will reply to your email.
Go to contact