Skip to main content
GET
Whether to recommend pausing a rollout

Authorizations

Authorization
string
header
required

A project token (apsio_pt_v0...), which reads exactly one project, or an OAuth access token issued for the API, which reads what its user can. Access tokens issued for Apsio's MCP server are accepted only from the MCP server itself.

Path Parameters

projectId
string
required
Pattern: ^[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$
Example:

"0192f3a4-0000-7000-8000-00000000a101"

Query Parameters

app_id
string
required

The app: iOS and Android differ almost everywhere.

Pattern: ^[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$
release
string
required

The release to check.

Required string length: 1 - 128
baseline
string

The release to compare it with. Default: the one before it in the app.

Required string length: 1 - 128
build
enum<string>

Which builds: release (the default), debug or all.

Available options:
release,
debug,
all

Response

The recommendation, or why there is none.

summary
string
required

Apsio's own words and numbers: no release name (D79).

verdict
enum<string>
required

consider_pausing: worse at the same age, for a person to weigh. no_recommendation: says nothing about whether the release is fine. too_few: under the minimum sample, so no verdict and no ratio.

Available options:
consider_pausing,
no_recommendation,
too_few
blocked
boolean
required

With no_recommendation: the release reads worse per install, and something else may explain it. limits says what.

subject
object
required
windows
object | null
required
evidence
object | null
required
sample
object
required
limits
object[]
required

What this answer could not see. text is in English.

would_clear_it
string | null
required
computed_at
string
required

RFC 3339 time in UTC with up to nanosecond precision.

Example:

"2026-10-04T10:15:00.431000000Z"