Skip to content

v25.0.0

Choose a tag to compare

@LaunchDarklyReleaseBot LaunchDarklyReleaseBot released this 23 Oct 17:48
· 2 commits to main since this release
Immutable release. Only release title and notes can be modified.

Added

  • Added optional 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}/variations
    • PATCH /api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}
    • PATCH /api/v2/projects/{projectKey}/ai-configs/{configKey}/variations
    • Config and variation read responses that return variation objects
  • Added Environment releases (beta) endpoints for starting, stopping, and viewing guarded and progressive releases on a flag in one environment. Starting a release runs release health checks and returns 400 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}/releases
    • POST /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases
    • GET /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases/{releaseId}
    • PUT /api/v2/projects/{projectKey}/flags/{flagKey}/environments/{environmentKey}/releases/{releaseId}/stop

Changed

  • Renamed the release pipelines schemas that start with Release 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 PipelineRelease
    • CreateReleaseInput to CreatePipelineReleaseInput
    • ReleaseAudience to PipelineReleaseAudience
    • ReleaserAudienceConfigInput to PipelineReleaseAudienceConfigInput
    • ReleaseProgression to PipelineReleaseProgression
    • ReleaseProgressionCollection to PipelineReleaseProgressionCollection
    • ReleaseStrategy to PipelineReleaseStrategy
    • ReleaseGuardianConfiguration to PipelineReleaseGuardianConfiguration
    • ReleaseGuardianConfigurationInput to PipelineReleaseGuardianConfigurationInput
  • Deprecated the release pipelines endpoints, and renamed the Release 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-pipelines
    • POST /api/v2/projects/{projectKey}/release-pipelines
    • GET /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}/releases
    • PUT /api/v2/projects/{projectKey}/flags/{flagKey}/release
    • PUT /api/v2/projects/{projectKey}/flags/{flagKey}/release/phases/{phaseId}
    • GET /api/v2/flags/{projectKey}/{flagKey}/release
    • PATCH /api/v2/flags/{projectKey}/{flagKey}/release
    • DELETE /api/v2/flags/{projectKey}/{flagKey}/release
  • Documented the AI tool and agent skill fields on the /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.
  • Added a 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.
  • Documented the AI model config fields on the /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.
  • Added explicit 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.
  • Moved the releases endpoints from the 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}/release
    • PUT /api/v2/projects/{projectKey}/flags/{flagKey}/release/phases/{phaseId}
    • GET /api/v2/flags/{projectKey}/{flagKey}/release
    • PATCH /api/v2/flags/{projectKey}/{flagKey}/release
    • DELETE /api/v2/flags/{projectKey}/{flagKey}/release
  • POST /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.
  • Corrected the holdouts API description. It previously said you can delete holdouts, but there is no delete endpoint. You can create, update, and end holdouts. To end a holdout, use the endHoldout instruction on PATCH /api/v2/projects/{projectKey}/environments/{environmentKey}/holdouts/{holdoutKey}. Request behavior is unchanged.
  • Removed the beta tag from the views API. These endpoints no longer require the LD-API-Version: beta header. Affected endpoints:
    • GET /api/v2/projects/{projectKey}/views
    • POST /api/v2/projects/{projectKey}/views
    • GET /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}
  • Documented 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}/variations
    • PATCH /api/v2/projects/{projectKey}/ai-configs/{configKey}/variations/{variationKey}
  • The 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.
  • The 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-capabilities
    • GET /api/v2/integration-configurations/{integrationConfigurationId}
    • GET /api/v2/integration-configurations/keys/{integrationKey}
  • Documented the 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.
  • Corrected the 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.
  • Documented the agent skill key format on 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.
  • Added the 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}
  • Exposed the new expiry field on access tokens in the public OpenAPI spec.
  • Corrected the documented granularity rule on 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".
  • Environment-grouped usage metadata now includes 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.
  • Updated POST /api/v2/projects/{projectKey}/environments/{environmentKey}/contexts/search and GET /api/v2/projects/{projectKey}/environments/{environmentKey}/contexts/{kind}/{key} to document current behavior:
    • Search results are sorted by context kind and key, return one item per context with its most recently reported attributes, and include contexts received within the last 30 days. Get contexts returns one item per application, with 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.
  • Updated the supported filter fields and operators for POST /api/v2/projects/{projectKey}/environments/{environmentKey}/contexts/search:
    • The id field is no longer supported.
    • The 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.
    • The q field accepts a single string and performs a prefix search across the context key and the name, firstName, lastName, and email attributes.
    • Attribute filter path segments must be simple names (letters, numbers, underscores, hyphens); reference nested attributes by joining segments with periods. JSON pointer escaping (~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.
  • Continuation tokens issued before this change are no longer valid on the contexts search and get endpoints. Requests that use one receive a 400; restart pagination from the first page.
  • Deprecated the 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/search
    • GET /api/v2/projects/{projectKey}/environments/{environmentKey}/contexts/{kind}/{key}
    • POST /api/v2/projects/{projectKey}/environments/{environmentKey}/context-instances/search
    • GET /api/v2/projects/{projectKey}/environments/{environmentKey}/context-instances/{id}
  • Clarified that 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.
  • On both Experimentation usage endpoints, an unrecognized projectKey now returns 400 instead of being ignored.
  • The 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}
  • Data Export destination 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}
  • Direct updates to AgentControl config variations now return 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}/variations
  • Deprecated the sdkType 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}.
  • Renamed the 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}
  • Renamed the 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}/release
    • PATCH /api/v2/flags/{projectKey}/{flagKey}/release
    • DELETE /api/v2/flags/{projectKey}/{flagKey}/release
    • PUT /api/v2/projects/{projectKey}/flags/{flagKey}/release
    • PUT /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.
  • Updated the 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.
  • Updated the 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.
  • Made expiry on the access token API resource public.
  • Clarified the descriptions of 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.
  • Moved 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.
  • Renamed the summary of 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.
  • Grouped the adaptive triggers endpoints under the new 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-triggers
    • GET /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggers
    • GET /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}/enable
    • POST /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggers/{triggerID}/disable
  • Updated the POST /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.
  • Updated the following query params in the GET /api/v2/projects/{projectKey}/sdk-keys endpoint:
    • Added viewKey and environmentKey values to filter
    • Added viewSummaries value to expand
  • Added the 422 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.
  • Release policies (beta) now accept at most 10 stages in 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-policies
    • PUT /api/v2/projects/{projectKey}/release-policies/{policyKey}
  • Limited 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-policies
    • PUT /api/v2/projects/{projectKey}/release-policies/{policyKey}

Added

  • Added the optional 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}
  • Added the Environment releases (beta) API tag and its documentation for listing, retrieving, starting, and stopping environment releases.
  • Added 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.
  • Added the optional top-level 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 for a source with gaps in those 24 hours. Both fields are omitted when the watermark is unavailable. The response schema is now DataExportSeriesListRep, which adds these two fields to SeriesListRep.
  • Added groupBy=destinationId to GET /api/v2/usage/data-export-events and GET /api/v2/usage/warehouse-export. Each destination series has destinationId and destinationName metadata, destinationViewable when a current destination matches. The destination's own kind is added as destinationKind only when the caller can view the destination; a kind requested with groupBy=destinationKind is returned unchanged. Warehouse series use the LaunchDarkly destination ID when a current destination the caller can view matches, and the warehouse provider's destination ID, named Unknown destination, when none matches. Destinations the caller cannot view are combined into one Unknown destination series with destinationId hidden and destinationViewable false, one per combination of the other requested dimensions. Streaming IDs absent from the current configuration are named Unknown historical destination. Usage without a destination ID is one series with an empty destinationId, named Unattributed.
  • Added the sort query parameter to GET /api/v2/teams. Supported fields are name and memberCount. Prefix a field with - to sort in descending order, and separate multiple fields with commas to break ties. When sort is omitted, teams are ordered by creation date, newest first, as before.
  • Added the optional destinationKind query parameter and groupBy=destinationKind to GET /api/v2/usage/data-export-events and GET /api/v2/usage/warehouse-export. Values preserve source configuration/product identity: Streaming Snowflake is snowflake, Warehouse Snowflake is snowflake-v2, and they remain distinct Destination Types. The Warehouse destinationKind filter accepts snowflake-v2, databricks, bigquery, redshift, clickhouse, and s3. It cannot be combined with the existing destination filter, which retains its five pre-existing values and excludes s3; groupBy=destination and response metadata remain unchanged.
  • Added the beta GET /api/v2/usage/data-export-events/projects/ranking endpoint for bounded Streaming project ranking with explicit UTC bounds, selected project keys, search, cursor pagination, deterministic rank ordinals, and per-source counts. It returns 409 instead of a partial ranking when any successful usage row in the requested range remains unattributed after resolution through the account's current environment-to-project mapping.
  • Added the optional includePending query parameter to GET /api/v2/integration-configurations/keys/{integrationKey}. When true, the response includes pending two-step setup drafts that were created using /setup but are not yet finalized. It defaults to false, which returns only active configurations, matching the previous behavior.
  • Added notifyTeamKeys to the approval request resource. It lists the team keys sent in notifyTeamKeys when the approval request was created or when reviewers were added, so you no longer have to track that separately. The members of those teams are resolved when the request is made and are listed in notifyMemberIds. Team membership may have changed since. The field is omitted for approval requests created without team keys, including every request created before this change. Affected endpoints:
    • GET /api/v2/approval-requests
    • POST /api/v2/approval-requests
    • GET /api/v2/approval-requests/{id}
    • PATCH /api/v2/approval-requests/{id}
    • POST /api/v2/approval-requests/{id}/apply
    • POST /api/v2/approval-requests/{id}/reviews
    • GET /api/v2/projects/{projectKey}/flags/{featureFlagKey}/environments/{environmentKey}/approval-requests
    • POST /api/v2/projects/{projectKey}/flags/{featureFlagKey}/environments/{environmentKey}/approval-requests
    • POST /api/v2/projects/{projectKey}/flags/{featureFlagKey}/environments/{environmentKey}/approval-requests-flag-copy
    • GET /api/v2/projects/{projectKey}/flags/{featureFlagKey}/environments/{environmentKey}/approval-requests/{id}
    • PATCH /api/v2/projects/{projectKey}/flags/{featureFlagKey}/environments/{environmentKey}/approval-requests/{id}
    • POST /api/v2/projects/{projectKey}/flags/{featureFlagKey}/environments/{environmentKey}/approval-requests/{id}/apply
    • POST /api/v2/projects/{projectKey}/flags/{featureFlagKey}/environments/{environmentKey}/approval-requests/{id}/reviews
  • Added isAgent and iconUrl to the app object on audit log entries. isAgent is true when the application acted on behalf of the member named in the same entry, so the two are co-authors of the change rather than the member acting alone. It is omitted for applications that act on their own. iconUrl is the application's icon, when it has one. Affected endpoints:
    • GET /api/v2/auditlog
    • GET /api/v2/auditlog/{id}
  • Added adaptive triggers endpoints, which automatically change a flag or AgentControl config's targeting when an observability alert fires
    • POST /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggers
    • GET /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggers
    • GET /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}/enable
    • POST /api/v2/projects/{projectKey}/environments/{environmentKey}/adaptive-triggers/{triggerID}/disable
  • Added optional sort and filter query parameters to GET /api/v2/projects/{projectKey}/ai-configs/model-configs. sort accepts version, tags, and usedIn. Prefix a field with - for descending order. filter accepts query, tags, maintainerId, and maintainerTeamKey.
  • Added the optional usedInCount property to the ModelConfig representation. It is the number of distinct AgentControl configs that reference the model. Affected endpoints:
    • GET /api/v2/projects/{projectKey}/ai-configs/model-configs
    • POST /api/v2/projects/{projectKey}/ai-configs/model-configs
    • GET /api/v2/projects/{projectKey}/ai-configs/model-configs/{modelConfigKey}
    • PATCH /api/v2/projects/{projectKey}/ai-configs/model-configs/{modelConfigKey}
    • GET /api/v2/projects/{projectKey}/ai-configs/model-configs/{modelConfigKey}/versions
  • Added the optional read-only unitAggregationField property to the metric representation, reporting the column whose distinct values are counted for count_distinct metrics. It was already returned on the ratio denominator representation; it is now also documented on the top-level metric representation, so it appears wherever a metric is returned. Affected endpoints:
    • GET /api/v2/metrics/{projectKey}
    • POST /api/v2/metrics/{projectKey}
    • GET /api/v2/metrics/{projectKey}/{metricKey}
    • PATCH /api/v2/metrics/{projectKey}/{metricKey}
    • GET /api/v2/flags/{projectKey}
    • POST /api/v2/flags/{projectKey}
    • GET /api/v2/flags/{projectKey}/{featureFlagKey}
    • PATCH /api/v2/flags/{projectKey}/{featureFlagKey}
    • POST /api/v2/flags/{projectKey}/{featureFlagKey}/copy
  • Added a read-only, relative _site/_links.site link to several resource representations, pointing at the resource's page in the LaunchDarkly UI so consumers can deep-link without hardcoding routes or hosts (environment-scoped where the UI page is). Added to the AgentControl (AI Config), AI tool, prompt snippet, and agent graph representations, and to the _links.site of project, experiment, and approval request representations. Also corrected a stale _site example on the metric representation. Affected endpoints:
    • GET /api/v2/projects, GET /api/v2/projects/{projectKey}
    • GET /api/v2/projects/{projectKey}/environments/{environmentKey}/experiments, GET /api/v2/projects/{projectKey}/environments/{environmentKey}/experiments/{experimentKey}
    • GET /api/v2/approval-requests, GET /api/v2/approval-requests/{id}
    • GET /api/v2/projects/{projectKey}/ai-configs, GET /api/v2/projects/{projectKey}/ai-configs/{configKey}
  • Added the optional member query parameter to GET /api/v2/auditlog, POST /api/v2/auditlog, and POST /api/v2/auditlog/counts. It filters results to entries for changes made by the specified member, and may be repeated to filter by multiple members.
  • Added the optional token query parameter to GET /api/v2/auditlog, POST /api/v2/auditlog, and POST /api/v2/auditlog/counts. It filters results to entries for changes made by the specified access token, and may be repeated to filter by multiple tokens.
  • Added the optional defaultVariationKey property to the request body of PATCH /api/v2/projects/{projectKey}/ai-configs/{configKey}. It sets which variation the config serves by default, so you no longer need to edit the underlying flag's default variation directly.
  • Added PUT /api/v2/projects/{projectKey}/environments/{environmentKey}/sdk-keys/{sdkKeyKey}/views endpoint for replacing the set of views associated with an SDK key. The viewKeys field is required. To remove all views from the SDK key, pass an empty viewKeys array. Supports expand=viewSummaries.
  • Added viewKeys field to request body of POST /api/v2/projects/{projectKey}/environments/{environmentKey}/sdk-keys endpoint.
  • Added the critical filter value to GET /api/v2/projects/{projectKey}/environments. filter=critical:true returns only environments marked as critical, and filter=critical:false returns only non-critical environments. It combines with the existing query and tags filter values.
  • Added the optional clientName property to the response of GET /api/v2/caller-identity. For an OAuth access token, it is the registered name of the OAuth client the token was issued to. It is omitted for other kinds of credentials, and when the client's name can't be resolved.
  • Added the optional audience and expiresAt properties to the response of GET /api/v2/caller-identity. For an OAuth access token, audience lists the resources (RFC 8707) the token was issued for, and expiresAt is when the token expires. audience is omitted for tokens issued without a resource, and both are omitted for other kinds of credentials.

Removed

  • Removed the context instances endpoints from the public API documentation:
    • POST /api/v2/projects/{projectKey}/environments/{environmentKey}/context-instances/search
    • GET /api/v2/projects/{projectKey}/environments/{environmentKey}/context-instances/{id}
    • DELETE /api/v2/projects/{projectKey}/environments/{environmentKey}/context-instances/{id}