Read your own policy API
With GET /v1/agents/{agentId}/policy, an agent reads its own modes, lists, and limits. There's no API to change them.
http
GET /v1/agents/{agentId}/policyNeeds the Self (agent:read_self) scope. It returns the same content as MCP get_my_policy; the MCP version also adds the default sending address (from_address).
| Field | Meaning |
|---|---|
policy_version | The current policy version. It goes up each time the owner changes something. |
outbound.mode | Outbound mode: all, allowlist, blocklist, or none. legacy for an old mode. |
outbound.allow | The outbound allowlist |
outbound.block_count | How many entries are on the outbound blocklist (not the entries themselves) |
inbound.mode · inbound.allow · inbound.block_count | The same for inbound |
limits | The limits that actually apply: per_minute, per_hour, per_day, recipients_per_message, org_daily_recipients |
plan | The organization's plan |
how_to_change | How to change the policy (the owner, in the console) |
If the policy is being changed at that moment, you get 409 policy_changing. Read it again shortly.
The agent can't change its policy. If it needs to email someone who's blocked, it asks the owner to add them to the allowlist.
Feedback on this page? Write to support@atmark.ai.