LaunchDarkly API Client for Python
Repo launchdarkly/api-client-python, read from GitHub 22 min ago. launchdarkly has not joined Take My Repo.
disabled and description to each rule in GET /api/v2/projects/{projectKey}/ai-configs/{configKey}/targeting and the PATCH response for the same path. A disabled rule doesn't match any contexts, so it doesn't serve its variation or rollout. As in the flags API, each field is left out when the rule doesn't set it.creationDate filter to GET /api/v2/approval-requests, for listing approval requests created before or after a time. It takes the before and after operators with Unix milliseconds or an RFC 3339 timestamp, and both are exclusive. For example, filter=creationDate after 1794000000000 lists the approval requests created after that time.tools to agent skills, for pinning AI tools to an agent skill. On create and update, each entry is a SkillToolRef with the tool's key and an optional version, and leaving version out pins the tool's latest version at the time of the request. In responses, each entry is a PinnedSkillToolRef, which always includes the pinned version. Tools are listed in the order they reach the agent, each tool key can appear once, a skill can pin tools only from its own project, up to 20 by default, and you need read access to each tool. On PATCH, leaving tools out keeps the current tools, and sending it replaces them all; a change to the tools or their order creates a new skill version. This field is being released gradually: until it is enabled for your account, requests that send tools return a 400, and responses leave it out. Affected endpoints:
POST /api/v2/projects/{projectKey}/ai-configs/skillsPATCH /api/v2/projects/{projectKey}/ai-configs/skills/{skillKey}GET /api/v2/projects/{projectKey}/ai-configs/skillsGET /api/v2/projects/{projectKey}/ai-configs/skills/{skillKey}GET /api/v2/projects/{projectKey}/ai-configs/skills/{skillKey}/versionsskills and skillsTotalCount to GET /api/v2/projects/{projectKey}/ai-tools/{toolKey}/references. skills lists every agent skill version that pins a version of the tool, old versions included, as ToolSkillReference entries with skillKey, skillVersion, and toolVersion, ordered by skill key, then newest skill version first. limit and offset page items and skills separately. This field is being released gradually alongside agent skill tools, and responses leave it out until it is enabled for your account.409 response to DELETE /api/v2/projects/{projectKey}/ai-tools/{toolKey}. A tool can't be deleted while any version of an agent skill pins it, and the error message names those skills. Removing the tool from a skill creates a new skill version, and older versions keep their pins, so delete the skill to free the tool.bearerToken to the POST /api/v2/webhooks request body and to PATCH /api/v2/webhooks/{id}. If set, LaunchDarkly sends it in an Authorization: Bearer header with each webhook request. The token is write-only and is never returned. To remove it, patch /bearerToken with an empty string.hasBearerToken to the webhook object. It is true when the webhook sends a bearer token, and omitted otherwise. Affected endpoints:
GET /api/v2/webhooksPOST /api/v2/webhooksGET /api/v2/webhooks/{id}PATCH /api/v2/webhooks/{id}_id to each rule in GET /api/v2/projects/{projectKey}/ai-configs/{configKey}/targeting. It is the rule ID to pass as ruleId in targeting instructions that change a rule, such as addShadowRequest. Previously the ID was only available from the flags API.assignmentSourceType to the experiment object. It is the source of assignment data for the experiment, one of launchdarkly, snowflake, databricks, redshift, bigquery, or clickhouse. It defaults to launchdarkly, and experiments created before warehouse-native assignment existed omit it. Affected endpoints:
GET /api/v2/projects/{projectKey}/experimentsGET /api/v2/projects/{projectKey}/environments/{environmentKey}/experimentsPOST /api/v2/projects/{projectKey}/environments/{environmentKey}/experimentsGET /api/v2/projects/{projectKey}/environments/{environmentKey}/experiments/{experimentKey}PATCH /api/v2/projects/{projectKey}/environments/{environmentKey}/experiments/{experimentKey}assignmentDataSource to the iteration object and to iteration request bodies. It holds the assignment data source key, the experimentId to match in that source, and the optional startTime and endTime of the assignment data window. Responses also include the source's display name. It is omitted for experiments that assign through flags. You can also set it through the assignmentDataSource field on the updateExperimentFields instruction.assignmentVariantId to each treatment, in both iteration request bodies and responses. It is the variant identifier in the warehouse-native assignment source that maps to that treatment.assignmentDataSource and assignmentVariantId:
POST /api/v2/projects/{projectKey}/environments/{environmentKey}/experimentsPOST /api/v2/projects/{projectKey}/environments/{environmentKey}/experiments/{experimentKey}/iterationsPATCH /api/v2/projects/{projectKey}/environments/{environmentKey}/experiments/{experimentKey}rollup query parameter on POST /api/v2/auditlog. Set it to 1 to return rolled-up audit log entries, which group related changes into one entry. The parameter already existed and its behavior is unchanged, but it was previously left out of the spec.sort query parameter to GET /api/v2/projects/{projectKey}/ai-configs/skills. Use sort=usedIn to order agent skills by the number of AgentControl configs that use them, fewest first, or sort=-usedIn for most first. Without sort, skills are still ordered by creation date, newest first.next link in GET /api/v2/auditlog, POST /api/v2/auditlog and the REPORT form of /api/v2/auditlog dropping rollup. When a request sets rollup=1, its next link now includes rollup=1, so later pages are rolled up like the first page. Previously only the first page was rolled up.purpose query parameter to GET /api/v2/projects/{projectKey}/environments/{environmentKey}/sdk-active. Set purpose to ai to check for AI SDK activity instead of any SDK. Any other value, or none, keeps the existing behavior.judgeScores to AgentControl config metrics. It lists the average score and sample count for every judge metric recorded in the time range, keyed by the judge's evaluation metric key ($ld:ai:judge:<suffix>), including custom judges and the built-in accuracy, relevance, and toxicity judges. It is omitted when no judge scores were recorded. The existing judgeAccuracy, judgeRelevance, and judgeToxicity fields are unchanged. Affected endpoints:
GET /api/v2/projects/{projectKey}/ai-configs/{configKey}/metricsGET /api/v2/projects/{projectKey}/ai-configs/{configKey}/metrics-by-variationanalysisUnit on each metric in environment release responses. It is the context kind used as the metric's analysis unit. It is the release's randomization unit unless the metric uses a decoupled unit of analysis. Releases started before analysis units were recorded omit analysisUnit, and those metrics are analyzed on the randomization unit. Affected endpoints:
GET /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releasesPOST /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releasesGET /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases/{releaseId}PUT /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases/{releaseId}/stopanalysisUnit on each metric in POST /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases. It sets the context kind used as the metric's analysis unit. Omit it to analyze on the randomization unit. A value must differ from the randomization unit and be one of the metric's units. Percentile metrics, ratio metrics, and metrics in a group cannot set it.No gonfalon changelog entries. See the API changes in the release PR.
classifiers on AgentControl config variations. Send a non-empty array to store classifier rules on a judge variation; messages must be empty. Send an empty array to clear stored classifier rules. Responses include classifiers only when rules are stored. Affected endpoints:
POST /api/v2/projects/{projectKey}/ai-configs/{configKey}/variationsPATCH /api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}PATCH /api/v2/projects/{projectKey}/ai-configs/{configKey}/variations400 with code health_checks_failed when they fail. originalVariationValue and targetVariationValue hold the flag's variation values, which can be any JSON type. These are beta endpoints and require the LD-API-Version: beta header:
GET /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releasesPOST /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releasesGET /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases/{releaseId}PUT /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases/{releaseId}/stopRelease so they are not confused with the environment releases schemas. This only changes the schema names in the OpenAPI spec and the type names in generated API clients. The JSON returned by the API is unchanged. Renamed schemas:
Release to PipelineReleaseCreateReleaseInput to CreatePipelineReleaseInputReleaseAudience to PipelineReleaseAudienceReleaserAudienceConfigInput to PipelineReleaseAudienceConfigInputReleaseProgression to PipelineReleaseProgressionReleaseProgressionCollection to PipelineReleaseProgressionCollectionReleaseStrategy to PipelineReleaseStrategyReleaseGuardianConfiguration to PipelineReleaseGuardianConfigurationReleaseGuardianConfigurationInput to PipelineReleaseGuardianConfigurationInputRelease pipelines (beta) tag to Release pipelines (deprecated). The endpoints still require the LD-API-Version: beta header, and request behavior is unchanged. Affected endpoints:
GET /api/v2/projects/{projectKey}/release-pipelinesPOST /api/v2/projects/{projectKey}/release-pipelinesGET /api/v2/projects/{projectKey}/release-pipelines/{pipelineKey}PUT /api/v2/projects/{projectKey}/release-pipelines/{pipelineKey}DELETE /api/v2/projects/{projectKey}/release-pipelines/{pipelineKey}GET /api/v2/projects/{projectKey}/release-pipelines/{pipelineKey}/releasesPUT /api/v2/projects/{projectKey}/flags/{flagKey}/releasePUT /api/v2/projects/{projectKey}/flags/{flagKey}/release/phases/{phaseId}GET /api/v2/flags/{projectKey}/{flagKey}/releasePATCH /api/v2/flags/{projectKey}/{flagKey}/releaseDELETE /api/v2/flags/{projectKey}/{flagKey}/release/api/v2/projects/{projectKey}/ai-configs/tools and /api/v2/projects/{projectKey}/ai-configs/skills endpoints. version, tags, and customParameters on AI tools, and name, description, tags, and version on agent skills, now have descriptions and examples. version is the value to pass in a variation's tools or skills list. The state field on AI Config variation responses is now documented as an enum with the values archived and published. Response and request behavior is unchanged.tags filter to GET /api/v2/projects/{projectKey}/ai-configs/skills. Use filter=tags anyOf ["tag-a", "tag-b"] to return agent skills whose latest version has at least one of the tags. It can be combined with the existing query and key filters.GET /api/v2/tags now accepts agent-skill as a kind, and returns agent skill tags. When no kind is set, the response now also includes agent skill tags./api/v2/projects/{projectKey}/ai-configs/model-configs endpoints. Field descriptions and examples now cover the Provider.model-id key format, the difference between key and the provider model id, global, isRestricted, version, tags, maintainer fields, token costs, and the freeform params and customParams objects. The restricted models request and response fields are also documented. Request behavior is unchanged.format values to numeric fields that previously had none: Go int64 fields now emit format: int64 and Go float64 fields now emit format: double. The JSON returned by the API is unchanged -- these fields already carried 64-bit values -- but generated API clients will now type them as 64-bit rather than defaulting to 32-bit. This fixes truncation of large values such as Unix millisecond timestamps in SeriesTimeSliceRep. 41 fields are affected, across the usage, flags, metrics, projects, and destinations endpoints.Releases (beta) tag to the Release pipelines (beta) tag, so they are grouped with the other release pipeline endpoints in the documentation. Removed the Releases (beta) tag. Affected endpoints:
PUT /api/v2/projects/{projectKey}/flags/{flagKey}/releasePUT /api/v2/projects/{projectKey}/flags/{flagKey}/release/phases/{phaseId}GET /api/v2/flags/{projectKey}/{flagKey}/releasePATCH /api/v2/flags/{projectKey}/{flagKey}/releaseDELETE /api/v2/flags/{projectKey}/{flagKey}/releasePOST /api/v2/projects/{projectKey}/environments/{environmentKey}/experiments now defaults methodology to frequentist when it is not provided. Multi-armed bandit experiments still default to bayesian. Requests that set methodology are unchanged.endHoldout instruction on PATCH /api/v2/projects/{projectKey}/environments/{environmentKey}/holdouts/{holdoutKey}. Request behavior is unchanged.LD-API-Version: beta header. Affected endpoints:
GET /api/v2/projects/{projectKey}/viewsPOST /api/v2/projects/{projectKey}/viewsGET /api/v2/projects/{projectKey}/views/{viewKey}PATCH /api/v2/projects/{projectKey}/views/{viewKey}DELETE /api/v2/projects/{projectKey}/views/{viewKey}POST /api/v2/projects/{projectKey}/views/{viewKey}/link/{resourceType}DELETE /api/v2/projects/{projectKey}/views/{viewKey}/link/{resourceType}GET /api/v2/projects/{projectKey}/views/{viewKey}/linked/{resourceType}GET /api/v2/projects/{projectKey}/view-associations/{resourceType}/{resourceKey}modelConfigKey on AgentControl config variations. Use the model config's key from GET /api/v2/projects/{projectKey}/ai-configs/model-configs, such as OpenAI.gpt-4.1. Do not use a custom model config's id, which is the provider model ID. A modelConfigKey that does not match a model config still returns a 400, and the error message now names the key. Affected endpoints:
POST /api/v2/projects/{projectKey}/ai-configs/{configKey}/variationsPATCH /api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}kindKey, kindKeys, and kinds filters on POST /api/v2/projects/{projectKey}/environments/{environmentKey}/contexts/search now accept at most 1000 values per request; larger lists are rejected with a 400. Previously an oversized list failed the whole search once the generated query exceeded the search backend's maximum query size. Split larger lists across multiple requests.dynamicOptions object on integration manifest form variables now also reports source, filters, and allowUnknownValue. This lets a dynamicEnum field draw its options from a named LaunchDarkly catalog instead of the integration's own endpoint, narrow them by a sibling form variable, and keep a saved value that is no longer in the catalog. Existing endpoint/parser dynamic options are unchanged. New read-only schemas include DynamicOptionsSource and DynamicOptionFiltersItems. Affected endpoints:
GET /api/v2/integration-capabilitiesGET /api/v2/integration-configurations/{integrationConfigurationId}GET /api/v2/integration-configurations/keys/{integrationKey}state field on PATCH /api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}. The field is now an enum with the values archived and published. The description states what each value does and when the API rejects it. An AI Config itself cannot be archived. Request behavior is unchanged.basePermissions description for custom roles. Omitting basePermissions on POST /api/v2/roles defaults to reader. The description previously said the default was no_access. Request behavior is unchanged.POST /api/v2/projects/{projectKey}/ai-configs/skills. The key field description now states that keys must match ^[a-z0-9][a-z0-9-]*$ (lowercase letters, numbers, or -, starting with a lowercase letter or number), can be at most 64 characters, and can't be a Windows reserved file name such as con or nul. The API rejects keys that break these rules with a 400.405 method not allowed response to flag trigger create and update. It is returned when the environment requires approval for the stored flag change and the caller cannot bypass that requirement. Affected endpoints:
POST /api/v2/flags/{projectKey}/{featureFlagKey}/triggers/{environmentKey}PATCH /api/v2/flags/{projectKey}/{featureFlagKey}/triggers/{environmentKey}/{id}expiry field on access tokens in the public OpenAPI spec.GET /api/v2/usage/evaluations/{projectKey}/{environmentKey}/{featureFlagKey}: granularity is selected by the length of the requested range (under two hours minutely, under two days hourly, two days or longer daily), not by the age of the data. Documented that minutely data is retained for 70 days and that a sub-two-hour range starting beyond that is returned at hourly granularity (with the range start widened to the containing hour), that ranges predating available data return an empty series list, and that when a flag has served many distinct variation values the least recently seen values may be aggregated into an additional series labeled "Other Variations".projectKey and projectName. On the /api/v2/usage endpoints, a series grouped by environmentId without projectId now also carries the key and name of the environment's current project. Series grouped by projectId are unchanged.POST /api/v2/projects/{projectKey}/environments/{environmentKey}/contexts/search and GET /api/v2/projects/{projectKey}/environments/{environmentKey}/contexts/{kind}/{key} to document current behavior:
applicationId populated.totalCount is an approximation and may be omitted if it is unavailable.associatedContexts is deprecated and always returns 1.applicationId on search result items is always an empty string; it is populated only on Get contexts items.POST /api/v2/projects/{projectKey}/environments/{environmentKey}/contexts/search:
id field is no longer supported.kinds and kindKeys fields match the context's own kind and kind:key. Their contains operator accepts exactly one value; notEquals is not supported.anyOf is supported only on the applicationId, key, kind, kindKey, kindKeys, and kinds fields. For attributes, use equals with an array value, which matches any element.q field accepts a single string and performs a prefix search across the context key and the name, firstName, lastName, and email attributes.~0, ~1) is no longer supported. A filter may include at most 10 attribute clauses. startsWith values are limited to 500 characters and matching is case-sensitive.GET /api/v2/projects/{projectKey}/environments/{environmentKey}/context-attributes/{attributeName} now requires a kind filter; requests without one are rejected with a 400.400; restart pagination from the first page.sort query parameter on the contexts and context instances search and get endpoints. These endpoints do not support sorting, and requests that include sort are rejected with a 400. Results are returned in the order the search backend produces them, which is not configurable. Omit the parameter. Affected endpoints:
POST /api/v2/projects/{projectKey}/environments/{environmentKey}/contexts/searchGET /api/v2/projects/{projectKey}/environments/{environmentKey}/contexts/{kind}/{key}POST /api/v2/projects/{projectKey}/environments/{environmentKey}/context-instances/searchGET /api/v2/projects/{projectKey}/environments/{environmentKey}/context-instances/{id}projectKey on GET /api/v2/usage/data-export-events can be used without environmentKey to scope Data Export events to whole projects. Previously, projectKey without environmentKey could return account-wide data.GET /api/v2/usage/data-export-events now accepts groupBy=projectId when one to 100 projectKey filters are provided. It returns 409 instead of a partial series when any successful usage row in the requested range remains unattributed after resolution through the account's current environment-to-project mapping; plain project filtering keeps its compatibility behavior.projectKey without environmentKey on GET /api/v2/usage/experimentation-events now filters usage to the selected projects' current environments. A selected project with no current environments returns an empty series instead of account-wide usage.projectKey without environmentKey on GET /api/v2/usage/experimentation-keys now filters usage by every selected project ID, including projects with no current environments, so retained materialized usage remains available instead of returning account-wide usage.projectKey now returns 400 instead of being ignored.redshiftSetupScripts field on the integration configuration resource is now also returned for active (finalized) Redshift native Experimentation configurations, rehydrated from the persisted configuration. It was previously present only on the two-step setup endpoint responses. The scripts are regenerated from the stored configuration on read and are not persisted, so the returned values are unchanged. Affected endpoints:
GET /api/v2/integration-configurations/{integrationConfigurationId}GET /api/v2/integration-configurations/keys/{integrationKey}GET, PATCH, and DELETE requests now return 404 when the destination does not belong to the environment in the request URL. Affected endpoints:
GET /api/v2/destinations/{projectKey}/{environmentKey}/{id}PATCH /api/v2/destinations/{projectKey}/{environmentKey}/{id}DELETE /api/v2/destinations/{projectKey}/{environmentKey}/{id}403 when a changed variation is in use in an environment that requires approval. A variation is in use when its backing flag variation is referenced by fallthrough, a targeting rule, an individual or context target, or a non-zero percentage rollout. Affected endpoints:
PATCH /api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}PATCH /api/v2/projects/{projectKey}/ai-configs/{configKey}/variationssdkType query parameter and the sdkType groupBy value on GET /api/v2/usage/service-connections. Both are still accepted. This endpoint reports server-side connections only. For client-side connections, use GET /api/v2/usage/streams/{source}.DependentMeasuredRolloutRep schema to DependentReleasePhaseRep. This only changes the schema name in the OpenAPI spec and the type name in generated API clients. The JSON returned by the API is unchanged. The schema appears in the guardedRollouts field of the metric resource. Affected endpoints:
GET /api/v2/metrics/{projectKey}POST /api/v2/metrics/{projectKey}GET /api/v2/metrics/{projectKey}/{metricKey}PATCH /api/v2/metrics/{projectKey}/{metricKey}DELETE /api/v2/metrics/{projectKey}/{metricKey}ReleasePhase schema to PipelineReleasePhase. This only changes the schema name in the OpenAPI spec and the type name in generated API clients; the JSON returned by the API is unchanged. Affected endpoints:
GET /api/v2/flags/{projectKey}/{flagKey}/releasePATCH /api/v2/flags/{projectKey}/{flagKey}/releaseDELETE /api/v2/flags/{projectKey}/{flagKey}/releasePUT /api/v2/projects/{projectKey}/flags/{flagKey}/releasePUT /api/v2/projects/{projectKey}/flags/{flagKey}/release/phases/{phaseId}POST /api/v2/projects/{projectKey}/views now converts the key in the request body to lowercase before it creates the view. The lowercase key is what the response returns, what the uniqueness check compares against, and what role policy resource specifiers must use. Views created before this change keep their original casing, and the other view endpoints continue to match keys exactly as stored.DELETE /api/v2/metrics/{projectKey}/{metricKey} description to state that a metric must be archived before it can be deleted. Added the 409 status conflict response returned when a metric cannot be deleted because it is not archived or is still in use by an experiment, guarded rollout, metric group, or release policy.POST /api/v2/members description to document that the teamKeys field requires the teams entitlement, which is available only on Enterprise plans. A request that includes a non-empty teamKeys field on a plan without that entitlement is rejected with a 403, and no members are created. Omitting teamKeys continues to work on every plan that allows member management.expiry on the access token API resource public.trackEvents (feature flag and flag rule), trackEventsFallthrough, and defaultTrackEvents (environment): enabling them sends detailed evaluation events, which is required for Data Export, but doesn't enable Data Export on its own.GET /api/v2/projects/{projectKey}/agent-optimizations/{optimizationKey}/runs and GET /api/v2/projects/{projectKey}/ai-tools/{toolKey}/references from the AI Configs tag to the AgentControl tag, so they are grouped with the other AgentControl endpoints in the documentation.GET /api/v2/projects/{projectKey}/agent-optimizations/{optimizationKey}/results from "List agent optimization runs" to "List latest agent optimization result per run". It previously shared a title with the /runs endpoint even though the two return different data: /results returns the most recent full result for each run of the current optimization version, while /runs returns a lightweight run summary per run across all versions.Adaptive triggers tag, so they now appear in their own section of the API documentation. Affected endpoints:
POST /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggersGET /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggersGET /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggers/{triggerID}PATCH /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggers/{triggerID}DELETE /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggers/{triggerID}POST /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggers/{triggerID}/enablePOST /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggers/{triggerID}/disablePOST /api/v2/projects/{projectKey}/environments/{environmentKey}/sdk-keys description to document that deleted SDK keys are retained for 180 days. During that period you cannot reuse the key of a deleted SDK key, even though the deleted key no longer appears in the list of SDK keys for the environment. A request that reuses such a key is rejected with a 400. This documents existing behavior; the endpoint is unchanged.GET /api/v2/projects/{projectKey}/sdk-keys endpoint:
viewKey and environmentKey values to filterviewSummaries value to expand422 unprocessable entity response to PATCH /api/v2/projects/{projectKey}/environments/{environmentKey}/experiments/{experimentKey}. It is returned when the instructions in the request are well formed, but cannot be applied to the experiment as it currently stands. Requests that succeed today are unaffected.guardedReleaseConfig.stages and progressiveReleaseConfig.stages by default. Some accounts have a higher limit, up to 20 stages, which is the schema's maxItems. Stages are also validated: each stage must have an allocation greater than 0 and greater than the previous stage's, at most 50% (50000) for guarded releases or 100% (100000) for progressive releases, and a durationMillis greater than 0. Requests that break these rules are rejected with a 400. Existing release policies can still be read, deleted, and applied, but a PUT that sends back more stages than the account's limit, or invalid stages, is rejected until the stages are fixed. Affected endpoints:
POST /api/v2/projects/{projectKey}/release-policiesPUT /api/v2/projects/{projectKey}/release-policies/{policyKey}scope.flagTagKeys on release policies (ReleasePolicyScope). A policy can have at most 1000 flag tag keys. Flag tag keys have the same limitations as flag tags: each key must only contain letters, numbers, -, _, or ., and be at most 64 characters. Requests that exceed these limits are rejected with a 400. Existing policies stay readable and deletable, but a PUT must drop or fix any stored keys that are over these limits. Affected endpoints:
POST /api/v2/projects/{projectKey}/release-policiesPUT /api/v2/projects/{projectKey}/release-policies/{policyKey}clickHouseSetupScript field to the integration configuration resource for ClickHouse native Experimentation. It holds the SQL grants the customer runs in ClickHouse, regenerated from the stored configuration, and never contains a password. Affected endpoints:
GET /api/v2/integration-configurations/{integrationConfigurationId}GET /api/v2/integration-configurations/keys/{integrationKey}Environment releases (beta) API tag and its documentation for listing, retrieving, starting, and stopping environment releases.environmentKey as a supported value for the sort query parameter on GET /api/v2/projects/{projectKey}/sdk-keys. It orders SDK keys by environment key ascending, then by creation time ascending within each environment, so each page is environment-contiguous. createdAt and -createdAt are unchanged.dataThrough and delayed fields to the responses of GET /api/v2/usage/data-export-events and GET /api/v2/usage/warehouse-export. dataThrough is the end of the latest hour that the source's usage pipeline has written across all accounts, as an RFC 3339 UTC timestamp. delayed is true when the source wrote every hour of the 24 hours ending at dataThrough and more than three hours have passed since dataThrough. It is always false