Skip to main content
PATCH
Update agent policy

Authorizations

Authorization
string
header
required

API token authentication using format <api token id>:<api client secret>

Path Parameters

agentId
string
required

System-generated unique agent identifier

Body

application/json

Partial update to an agent's policy. Only provided fields will be updated; omitted fields retain their current values.

permissions
enum<string>[]

Updated list of permissions. Replaces the entire permissions list when provided.

Permission granted to an agent that determines what actions it can perform. VIEW_TRANSACTIONS: Can list and retrieve transactions and account balances. CREATE_QUOTES: Can create quotes. EXECUTE_QUOTES: Can execute quotes. MANAGE_EXTERNAL_ACCOUNTS: Can create and manage external accounts. RECEIVE_FUNDS: Can create deposit targets for receiving funds. MANAGE_IDENTITY: Can read and update customer identity information and start verification. MANAGE_CARDS: Can read, tighten, and freeze its cards. ISSUE_CARDS: Can issue one standard virtual card for itself while it holds no live standard card, and single-use and recurring purchase cards where the platform enables them. USE_STORED_CARDS: Can list the customer's stored cards it is allowed to use and ask for a single-use link to one's details. Never granted by default. MANAGE_SETTLEMENT_RULES: Can propose settlement rules that split the customer's incoming payments into its Embedded Wallet, read and delete the rules it created, and cancel their held executions.

Available options:
VIEW_TRANSACTIONS,
CREATE_QUOTES,
EXECUTE_QUOTES,
MANAGE_EXTERNAL_ACCOUNTS,
RECEIVE_FUNDS,
MANAGE_IDENTITY,
MANAGE_CARDS,
ISSUE_CARDS,
USE_STORED_CARDS,
MANAGE_SETTLEMENT_RULES
defaultExecutionMode
enum<string>

Execution mode controlling whether agent actions require human approval. AUTO: The agent can execute actions autonomously without explicit approval. APPROVAL_REQUIRED: All agent actions require explicit human approval before execution.

Available options:
AUTO,
APPROVAL_REQUIRED
spendingLimits
object

Partial update to spending limits. Only provided fields will be updated; omitted fields retain their current values.

accountRestrictions
object

Optional restrictions that limit the agent to specific accounts or override policy per account.

approvalThresholds
object

Thresholds that force approval for high-value transactions, overriding the default execution mode. The threshold currency must equal the policy's spending limits currency. When a transaction is denominated in a different currency, Grid converts it to the policy currency at evaluation time.

Response

Agent policy updated successfully

A programmatic agent with scoped permissions and a spending policy, used to automate payment workflows.

id
string
required

System-generated unique identifier for the agent.

Example:

"Agent:019542f5-b3e7-1d02-0000-000000000001"

name
string
required

Human-readable name for the agent.

Example:

"Payroll Automation Agent"

customerId
string
required

The ID of the customer this agent operates on behalf of.

Example:

"Customer:019542f5-b3e7-1d02-0000-000000000001"

isPaused
boolean
required

Whether the agent is currently paused. Paused agents cannot initiate any actions.

Example:

false

isConnected
boolean
required

Whether the agent has been installed and connected (i.e., its device code has been redeemed).

Example:

true

policy
object
required

Policy governing what an agent can do, how it executes actions, and its spending boundaries.

usage
object
required

Real-time counters tracking the agent's spending and transaction activity against its policy limits.

createdAt
string<date-time>
required

Creation timestamp.

Example:

"2025-07-21T17:32:28Z"

updatedAt
string<date-time>
required

Last update timestamp.

Example:

"2025-07-21T17:32:28Z"

actingFor
string

The platform's reference for the person this agent acts for, as given when the agent was created. Absent when none was given.

Example:

"member_8f2c1a"

spendingPoolId
string

The spending pool the agent spends under, in addition to its own policy. Absent when the agent is in no pool.

Example:

"SpendingPool:019542f5-b3e7-1d02-0000-000000000007"