Skip to main content Crate distri_types Copy item path Source pub use workflow_triggers::WebhookAuth ;pub use workflow_triggers::WebhookResponse ;pub use workflow_triggers::WorkflowTrigger ;pub use api_error::ApiError ;pub use api_error::ApiResult ;pub use configuration::AgentConfig ;pub use tool_result_store::CacheCheck ;pub use tool_result_store::ContentFormat ;pub use tool_result_store::ContentReplacementState ;pub use tool_result_store::FILE_UNCHANGED_STUB ;pub use tool_result_store::FileReadCache ;pub use tool_result_store::MAX_TOOL_RESULT_CHARS ;pub use tool_result_store::MAX_TOOL_RESULTS_PER_MESSAGE_CHARS ;pub use tool_result_store::PERSIST_THRESHOLD_BYTES ;pub use tool_result_store::PREVIEW_SIZE_BYTES ;pub use tool_result_store::PersistedToolResult ;pub use tool_result_store::Preview ;pub use tool_result_store::ReplacementDecision ;pub use auth ::*;pub use events ::*;pub use tenant_context ::*;pub use tool ::*;pub use integration ::*;pub use filesystem ::*;pub use skill ::*;pub use todos ::*;pub use ui_tool_render ::*;pub use ui_tool_renderers ::*;pub use models ::*;a2a a2a_converters api api_error ApiError — the single error type that every distri service returns.auth browser channel_commands Channel-facing rendering types. A WorkflowAgent bound to a bot
uses these for slash command declarations (now on
crate::WorkflowTrigger::Slashdocs/specs/workflow-channel-commands.md. channels Shared channel + bot types. Consumed by distri-cloud (server, stores,
handlers) and distri-gateway (webhook adapters) so the two halves speak
the same vocabulary. configuration connections context dynamic_tool events filesystem http_request Typed request/response structs for the HTTP request proxy. integration invocation The unified sub-agent invocation model. mock_tool Typed config for the mock dynamic-tool factory. model_catalog Provider catalog — the section-grouped file format for provider/model
definitions, plus loaders. models Core model, provider, and audio types used across the entire system. prompt resolve skill stores tenant_context thinking Branded thinking/loading phrases shared between CLI and UI. todos tool tool_result_store Two-level tool result persistence and file read caching. ui_tool_render ui_tool_renderers workflow_triggers Unified workflow trigger taxonomy. AdditionalAttributes Additional attributes for thread/task metadata. AdditionalParts Structure for managing additional user message parts
This allows control over how parts are added and whether artifacts should be expanded AgentContextSize AgentPlan AgentStrategy Unified Agent Strategy Configuration AvailableSkill A reference to a stored skill that an agent can load on demand CompactionSummary Summary produced by semantic compaction of older scratchpad entries ContextBudget Tracks token usage by component for context optimization. ContextSize ContextUsage CreateThreadRequest DistriConfig Configuration for the Distri client. ExecutionHistoryEntry Enriched execution history entry that includes context metadata ExecutionResult Execution result with detailed information ExecutorContextMetadata Metadata sent by clients (CLI, browser SDK, etc.) alongside A2A messages.
Deserialized by the server to configure execution context.
This is the canonical schema — all clients should serialize this struct. ExternalTool External tool that delegates execution to the frontend HookContext HookMutation InlineHookRequest InlineHookResponse LLmContext LlmDefinition McpDefinition McpServerHandle A pool-ready handle for one MCP server: identifier + transport + already-
resolved authorization headers. McpServerMetadata McpSession McpToolConfig Configuration for tools from an MCP server Message MockOrchestrator ModelInfo A model entry within a provider ModelSettings Model settings configuration.
A ModelSettings always has a valid model string.
Use Option<ModelSettings> when no model is configured yet. ModelSettingsInner Optional/defaultable model parameters. Split from ModelSettings so callers
can construct ModelSettings { model: "...", ..Default::default() } easily
via the inner field having Default. OrchestratorRef Reference wrapper for orchestrator that allows late binding PartMetadata Metadata for individual message parts.
Used to control part behavior such as persistence and rendering. PlanConfig Configuration for planning operations PlanStep Plan step for execution ProviderModels Models grouped by provider, with configuration status ProviderModelsStatus Provider models with configuration status (returned by API) ProviderSecretDefinition Defines the secret requirements for a provider ReflectionConfig Reflection configuration ReplanningConfig Replanning configuration ResourceLink MCP resource reference. Mirrors the shape an MCP server returns under
tools/call’s content[].resource, plus the _meta blob that hosts use
to discover MCP-Apps UI capabilities (_meta.ui.resourceUri). ScratchpadEntry Entry for scratchpad formatting SecretKeyDefinition Defines a single secret key requirement ServerMetadataWrapper SkillContextEntry Skill content re-injected after compaction to preserve agent instructions StandardDefinition Agent definition - complete configuration for an agent Task TaskEvent Thread ThreadSummary TokenUsage Token usage breakdown from an LLM call. ToolCall ToolDefinition ToolResponse ToolsConfig TraceContext Inbound distributed-trace context, propagated by callers so an agent run
(and all of its descendant spans) nests under a remote parent trace. UpdateThreadRequest UserMessageOverrides Action Action can be either a tool call or an LLM call AdditionalPartsInstruction Instruction for how to handle additional parts AgentError CodeLanguage ExecutionKind ExecutionMode Execution mode - tools vs code ExecutionStatus ExecutionType Execution strategy types FileType HookKind McpClientTransport Client-side transport configuration for connecting to an MCP server. McpServerType MemoryKind MessageRole ModelProvider OpenAiApiFormat Which OpenAI-family API format to use when talking to the LLM. Part PartDefinition ReasoningDepth Depth of reasoning for planning ReflectionDepth Depth of reflection ReflectionTrigger When to trigger reflection ReplanningTrigger When to trigger replanning ResolvedOpenAiApiFormat Resolved (non-Auto) API format RuntimeMode Runtime environment in which the agent is executing.
Determines which built-in agent variants and tools are available. ScratchpadEntryType Type of scratchpad entry - only for Thought/Action/Observation tracking TaskMessage TaskStatus Lifecycle state of a task. Snake-case in JSON to match the
tasks.status schema column (“pending” / “running” /
“input_required” / “completed” / “failed” / “canceled”) and the
SDK’s typed AgentTaskStatus enum. ToolCallFormat Supported tool call formats ToolCallStatus ToolDeliveryMode How tools are delivered to the LLM in the prompt. ToolResultWithSkip TransportType CORE_TOOLS Tools that always get full schemas, never deferred.
These are the most commonly used tools that agents need immediately. DEFAULT_DEFERRED_THRESHOLD Default threshold: defer tools when total count exceeds this. DEFAULT_EXTERNAL_TOOL_TIMEOUT_SECS Default timeout for external tool execution in seconds VALID_BUILTIN_TOOLS Tools configuration for agents
Canonical list of valid builtin tool names. OrchestratorTrait Trait for workflow runtime functions (session storage, agent calls, etc.) ServerTrait default_agent_version default_transport_type from_tool_results lookup_provider_test_config Look up the provider-test override for a registered provider. The
/v1/providers/test handler calls this and falls back to its default
GET /models probe when None is returned. Reads from the merged
layered registry (built-in basics + deployment extensions). parse_agent_markdown_content register_provider_extensions Register deployment-owned provider/model definitions — layer 2 of the
provider registry. Call once, at process startup, before any
provider/model catalog is served. validate_parameters validate_plugin_name Validate plugin name follows naming conventions
Plugin names must be valid identifiers. At most one ‘/’ is allowed for workspace namespacing (e.g. ‘workspace/agent’). BuilderFn PartsMetadata Mapping of part indices to their metadata.
Used in message metadata to specify per-part behavior.