Agent identity
Agent name, version, owner, primary model or platform, and intended business workflow.
Before submitting an agent
Use this checklist to prepare a safe, useful first request. You do not need to send private data, secrets, or executable files to begin.
Preparation checklist
The more clearly these are defined, the stronger and faster the initial review scope can be.
Agent name, version, owner, primary model or platform, and intended business workflow.
What the agent may do, what it must never do, and where human approval is required.
What the agent can remember, who controls memory, and how memory can be corrected or deleted.
APIs, files, web access, automations, wallets, account actions, or no tools.
When the agent should pause, refuse, ask for clarification, or hand off to a human.
Customer-facing pressure, emotional sensitivity, regulated domains, payment risk, privacy risk, or autonomy risk.
Safe first evidence
Ready to submit
A useful first request names the workflow, the intended users, the allowed actions, known risks, and what kind of review you want.
What real workflow should the agent support?
What behavior worries you most: memory, tools, escalation, emotional pressure, overclaiming, or reliability?
Snapshot, Behavior Review, Continuous Passport, or Metamorphosis scoping.
What can be reviewed safely without exposing private data?
Checklist boundary
Readiness to submit is not review approval, Passport status, regulatory approval, legal certification, guaranteed safety, or a claim about consciousness, sentience, personhood, emotion, or inner experience.