Resource
ai_adapter
A configured AI adapter — an LLM provider or ML backend the tenant can run inference against. The API key is withheld; everything else is readable.
Configuration
nameString!RequiredDisplay name (required)
enabledBoolean!OptionalWhether the adapter is enabled (optional, defaults to true)
configurationAIAdapterConfiguration!RequiredThe adapter's configuration; exactly one non-null branch determines the adapter type.
Input types
Fields for the structured values used in this resource’s configuration.
AIAdapterConfigurationInput
The adapter's configuration. Supply exactly one non-null branch to select the adapter type. Mirrors the corresponding output configuration field for field apart from the API key, which is write-only and never returned — so a configuration read from the API can be sent back unchanged, and omitting the key keeps the stored one.
Set exactly one field to a non-null value.
openaiOpenAIConfigurationInputOptionalOpenAI configuration. Set this branch to select OpenAI.
claudeClaudeConfigurationInputOptionalClaude configuration. Set this branch to select Claude.
xgboostXGBoostConfigurationInputOptionalXGBoost configuration. Set this branch to select XGBoost.
OpenAIConfigurationInput
Configuration for OpenAI adapter
apiKeyStringOptionalOpenAI API key. Optional on update: omit it (or send an empty string) to keep the stored key — it is never readable, so a client has nothing to send back. Required to create an adapter.
organizationIdStringOptionalOrganization ID (optional)
baseURLStringOptionalBase URL for custom endpoints (optional)
defaultChatModelStringOptionalDefault model for chat (optional)
defaultCodeModelStringOptionalDefault model for code (optional)
defaultCompletionModelStringOptionalDefault model for completions (optional)
ClaudeConfigurationInput
Configuration for Claude adapter
apiKeyStringOptionalClaude API key. Optional on update: omit it (or send an empty string) to keep the stored key — it is never readable, so a client has nothing to send back. Required to create an adapter.
baseURLStringOptionalBase URL for custom endpoints (optional)
defaultChatModelStringOptionalDefault model for chat (optional)
defaultCodeModelStringOptionalDefault model for code (optional)
defaultCompletionModelStringOptionalDefault model for completions (optional)
XGBoostConfigurationInput
Configuration for XGBoost adapter
maxDepthIntOptionalMaximum tree depth (optional; the trainer's default is 6)
etaFloatOptionalLearning rate (optional; the trainer's default is 0.3)
roundsIntOptionalNumber of boosting rounds (optional; the trainer's default is 100)
Read-only values
Show values
idAIAdapterID!Read onlyUnique identifier
computed by the platform
tenantIdTenantID!Read onlyTenant this adapter belongs to
computed by the platform
capabilitiesAIAdapterCapabilities!Read onlyCapabilities this adapter supports
computed by the platform
hasApiKeyBoolean!Read onlyWhether a non-empty API key is stored. The key itself is never returned — not even masked, since a mask a client could send back would be indistinguishable from the real thing. Always false for adapter types that do not authenticate (XGBOOST).
computed by the platform
credentialsUpdatedAtStringRead onlyWhen the stored API key last changed (ISO8601). Null when no key is set, and also null for a key last written before the platform recorded this — so null means "unknown", never "never changed". Deliberately a timestamp and not a hash: it says when the credential changed, never what it is, so it lets an operator spot an unexpected rotation without disclosing anything.
computed by the platform
createdAtString!Read onlyCreation timestamp (ISO8601)
computed by the platform
updatedAtString!Read onlyLast update timestamp (ISO8601)
computed by the platform
Lifecycle
The CLI can create this resource. The platform assigns its id.
When replacement is needed, the existing resource is removed before its replacement is created.