Neon API (unofficial showcase)
POST/projects/{project_id}/branches/{branch_id}/logs/query

Query branch logs

Returns logs emitted by services running on the specified branch, ordered by timestamp according to sort_order.

All supplied filters are combined with AND: a record is returned only when it matches every filter. minimum_severity and severity_text are independent filters, so setting both requires a record to clear the severity floor and match the exact severity text.

Supply logql instead of the structured filters to run a raw LogQL expression. Combining it with any structured filter is rejected rather than silently ignored; limit, sort_order, and the time window still apply, because those bound the query rather than form part of the expression.

Give the window either as since — a duration ending at end_time, or at the current time when end_time is omitted — or as an explicit start_time. Supplying both is rejected.

A single response holds at most 1,000 records. When is_truncated is true, pass the returned next_cursor back as cursor to fetch the next page, repeating the time range and every filter unchanged.

If no time range is supplied, the query covers the previous hour. The maximum supported time range is seven days. end_time is exclusive.

Note: This endpoint is currently in Private Beta.

  • IdempotentThe SDK sends Idempotency-Key, so a retried request is only applied once.

2 parameters · 14 body fields
project_idstringrequired
The Neon project ID
branch_idstringrequired
The Neon branch ID
sinceobjectoptional
Length of the query window, ending at `end_time` or at the current time when `end_time` is omitted. Mutually exclusive with `start_time`. Prefer this over computing absolute bounds when the caller only means "the last hour".
start_timestringoptional
Inclusive beginning of the query window. Mutually exclusive with `since`. Defaults to one hour before `end_time`, or one hour before the current time when both bounds are omitted.
end_timestringoptional
Exclusive end of the query window. Defaults to the current time.
limitintegeroptional
Maximum number of log records to return per page.
Default:100
cursorstringoptional
Opaque pagination cursor returned as `next_cursor` by a previous call. Resume the query after the last record of the previous page, repeating the time range and every filter unchanged.
sort_orderstringoptional
Order matching records by timestamp. `desc`, the default, returns the newest records first.
Allowed:ascdescDefault:desc
sourcestringoptional
The Neon service that emitted the log record.
Allowed:functionstoragepg_endpoint
service_namestringoptional
Match the OpenTelemetry `service.name` resource attribute exactly.
scope_namestringoptional
Match the OpenTelemetry instrumentation scope name exactly.
minimum_severitystringoptional
An OpenTelemetry severity level. A minimum severity includes every higher level in this order: `trace`, `debug`, `info`, `warn`, `error`, `fatal`.
Allowed:tracedebuginfowarnerrorfatal
severity_textstringoptional
Match the OpenTelemetry severity text exactly.
body_containsstringoptional
Match records whose rendered `message` contains this case-sensitive substring. Records with a structured body are matched against their JSON rendering, so the substring meets JSON syntax rather than prose: a bare key name such as `operation` matches every record carrying that key, and `http_status: 200` matches none, because the rendering contains `"http_status":200` with no space.
trace_idstringoptional
Match records associated with this OpenTelemetry trace ID. W3C Trace Context defines a trace ID as 32 lowercase hex digits, and that is what is stored, so an uppercase value is rejected rather than silently matching nothing.
logqlstringoptional
Escape hatch for selections the structured filters cannot express: a raw LogQL expression, evaluated against this branch's log stream. Only stream selectors and line filters are accepted — no aggregations and no parser stages. Supplying this alongside any structured filter is rejected with `conflicting_filters` rather than silently ignoring one of them. `limit`, `sort_order`, and the time window still apply. This field passes the underlying query language through to the caller, so unlike the rest of this contract it may change as that backend changes. Prefer the structured filters where they suffice.

4 status codes
200Logs matching the supplied filters
logsarray<object>required
next_cursorstringoptional
Pagination cursor to pass as `cursor` on the next request. Empty when the response is not truncated.
is_truncatedbooleanrequired
True when more records matched than were returned.
400The query could not be served as written. The body is always `ProjectBranchLogsInvalidQuery` — see `reason` for the exact cause.
codestringrequired
messagestringrequired
reasonstringrequired
Machine-readable reason why the request was rejected: - `time_range_too_large`: the requested window spans more than seven days. - `invalid_time_range`: `end_time` is not after `start_time`. - `conflicting_time_range`: both `since` and `start_time` were supplied. - `invalid_cursor`: the supplied `cursor` is malformed, expired, or was issued for a different query. - `unknown_field`: the requested `field_name` is not one of the fields the log fields endpoint reports for this branch. - `invalid_logql`: the supplied `logql` expression does not parse, or uses a construct this endpoint does not accept. - `conflicting_filters`: `logql` was supplied alongside one or more structured filters. Use one or the other.
Allowed:time_range_too_largeinvalid_time_rangeconflicting_time_rangeinvalid_cursorunknown_fieldinvalid_logqlconflicting_filters
404Logs are not available for this branch, or the project/branch was not found. The body is always `ProjectBranchLogsNotAvailable` — see `reason` for the exact cause.
codestringrequired
messagestringrequired
reasonstringrequired
Machine-readable reason why logs cannot be read: - `branch_not_found`: the project or branch does not exist, or the caller does not have access to it. - `telemetry_not_enabled`: the branch exists but is not collecting telemetry, so it has no logs to serve.
Allowed:branch_not_foundtelemetry_not_enabled
defaultGeneral Error. The request may or may not be safe to retry, depending on the HTTP method, response status code, and whether a response was received. - If no response is returned from the API, a network error or timeout likely occurred. - In some cases, the request may have reached the server and been successfully processed, but the response failed to reach the client. As a result, retrying non-idempotent requests can lead to unintended results. The following HTTP methods are considered non-idempotent: `POST`, `PATCH`, `DELETE`, and `PUT`. Retrying these methods is generally **not safe**. The following methods are considered idempotent: `GET`, `HEAD`, and `OPTIONS`. Retrying these methods is **safe** in the event of a network error or timeout. Any request that returns a `503 Service Unavailable` response is always safe to retry. Any request that returns a `423 Locked` response is safe to retry. `423 Locked` indicates that the resource is temporarily locked, for example, due to another operation in progress.
request_idstringoptional
Unique identifier for the request, useful for debugging. You can set this value manually by including an `X-Request-ID` header in the request. If not provided, the value will be generated automatically.
codestringrequired
Machine-readable code classifying the error type. See `message` for a human-readable explanation.
Default:
messagestringrequired
Error message

Error handling

A 400 is returned: The query could not be served as written. The body is always ProjectBranchLogsInvalidQuery — see reason for the exact cause. A 404 is returned: Logs are not available for this branch, or the project/branch was not found. The body is always ProjectBranchLogsNotAvailable — see reason for the exact cause. Any other status is an error: General Error.

The request may or may not be safe to retry, depending on the HTTP method, response status code, and whether a response was received.

  • If no response is returned from the API, a network error or timeout likely occurred.
  • In some cases, the request may have reached the server and been successfully processed, but the response failed to reach the client. As a result, retrying non-idempotent requests can lead to unintended results.

The following HTTP methods are considered non-idempotent: POST, PATCH, DELETE, and PUT. Retrying these methods is generally not safe. The following methods are considered idempotent: GET, HEAD, and OPTIONS. Retrying these methods is safe in the event of a network error or timeout.

Any request that returns a 503 Service Unavailable response is always safe to retry.

Any request that returns a 423 Locked response is safe to retry. 423 Locked indicates that the resource is temporarily locked, for example, due to another operation in progress.