Get a policy entry
GET /api/policy-entries/{entryId}
Returns one policy entry by its ID, whatever its status: its topic, rule text, conditions, status and source.
Authorizations
Section titled “Authorizations ”Parameters
Section titled “ Parameters ”Path Parameters
Section titled “Path Parameters ”The unique identifier of the policy entry to read.
Responses
Section titled “ Responses ”The policy entry.
A policy entry defines a rule that governs how the AI should behave in specific situations. Unlike KB entries (which provide factual answers), policies specify constraints, guardrails, and behavioral guidelines. For example, ‘Never offer more than 10% discount without manager approval’ or ‘Always ask for an order number before looking up shipping status’. Policies can be created manually or proposed by the learning loop.
object
Unique identifier for the policy entry.
Example
pe1a2b3c-5678-9abc-def0-1234567890abThe tenant this policy belongs to.
Example
a0b1c2d3-4567-89ab-cdef-0123456789abThe workspace this policy is scoped to.
Example
w1a2b3c4-5678-9abc-def0-1234567890abThe topic or domain this policy applies to. Used by the AI to determine which policies are relevant to the current conversation.
Example
discount_authorizationThe policy rule in natural language. This is the instruction the AI follows.
Example
Never offer a discount above 10% without creating an attention item for manager approval. For discounts up to 10%, apply automatically if the customer has placed 3 or more orders.Optional structured data stored with the policy, as free-form JSON. The Policy Agent receives it together with the rule text when it retrieves the policy.
Example
{ "min_order_value": 500, "customer_tier": "gold"}How the policy came to be. manual is a policy written by a person, approved when it is created. extracted is a proposal the learning loop drafted from a resolved conversation. The API stores the value sent, and a policy created with a value other than manual starts as pending_review.
Example
manualConfidence score between 0 and 1, set by the learning loop when proposing policies. Null for manually created policies.
Example
0.85Where the policy stands, with the values of a knowledge base entry. approved: in use, the assistant applies it. pending_review: created for review, through the API with a sourceType other than manual; it is not applied until it is approved. proposed: drafted by oHallo, for example by the learning loop from a resolved conversation; it is not applied until it is approved, and a proposal that changes an existing policy is applied to that policy when it is approved. rejected: declined in review. archived: no longer in use, kept for the record.
Example
approvedTimestamp when the policy was created.
Example
2026-02-05T09:00:00.000ZTimestamp when the policy was last updated.
Example
2026-03-18T11:30:00.000ZNo policy entry exists with the given ID in the authenticated tenant, or its workspace is outside the API key’s brand or workspace restriction.
Standard error response returned by all endpoints on failure.
object
A human-readable error message describing what went wrong.
Example
Conversation not foundA machine-readable reason on some refusals, for example ‘key_lock’ when an API key’s brand or workspace restriction does not cover the action.
Example
key_lockOn a 403 for an API key without a scope the endpoint requires: the scopes the key lacks.
Example
[ "mcp:write"]On a 409 with code duplicate_question or duplicate_rule_text: the knowledge base entry or policy that already holds the question or rule text.
Example
11223344-5566-4788-99aa-bbccddeeff00On a 422 about one MCP connection a request names, from an assignment or an agent create or update (a connection that does not exist, one listed twice, or one that does not declare a tool named): the connection, as the request named it.
On the same 422: the tool names the connection does not declare.
Example
[ "get_orders"]On the same 422: the tools the connection declares.
Example
[ "get_order", "search_orders"]