Flyte SDK

These are the docs for Flyte SDK version 2.0

Flyte is the core Python SDK for the Union and Flyte platforms.

Directory

Classes

Class	Description
`flyte.AsyncFunctionTaskTemplate`	A task template that wraps an asynchronous functions.
`flyte.Backoff`	Exponential backoff policy applied between user retries.
`flyte.BaseCheckpoint`	Base type for task checkpoint helpers.
`flyte.Cache`	Cache configuration for a task.
`flyte.Checkpoint`	Checkpoint helper using `flyte.io.File` for all checkpoint blob I/O (load/save, async and sync).
`flyte.ConditionWebhook`	Webhook configuration for a condition notification.
`flyte.Cron`	Cron-based automation schedule for use with `Trigger`.
`flyte.Device`	Represents a device type, its quantity and partition if applicable.
`flyte.Documentation`	This class is used to store the documentation of a task.
`flyte.Environment`	Base class for execution environments, shared by `TaskEnvironment` and `AppEnvironment`.
`flyte.FixedRate`	Fixed-rate (interval-based) automation schedule for use with `Trigger`.
`flyte.Image`	Container image specification built using a fluent, two-step pattern.
`flyte.ImageBuild`	Result of an image build operation.
`flyte.OnArtifact`	Artifact-based automation for use with `Trigger`: fire a run whenever a new version of the named artifact is created.
`flyte.PodTemplate`	Custom PodTemplate specification for a Task.
`flyte.Resources`	Resources such as CPU, Memory, and GPU that can be allocated to a task.
`flyte.RetryStrategy`	Retry strategy for a task.
`flyte.ReusePolicy`	Configure a task environment for container reuse across multiple task invocations.
`flyte.Secret`	Secrets are used to inject sensitive information into tasks or image build context.
`flyte.TaskEnvironment`	Define an execution environment for a set of tasks.
`flyte.TaskTemplate`	Task template is a template for a task that can be executed.
`flyte.Timeout`	Timeout bounds for a task.
`flyte.Trigger`	Specification for a trigger that can be associated with any Flyte task.
`flyte.ai.agents.AccessDenied`	Raised when a write targets a read-only or reserved prefix.
`flyte.ai.agents.Agent`	A flyte-native tool-use agent harness.
`flyte.ai.agents.AgentEvent`	Lightweight event emitted by the agent loop.
`flyte.ai.agents.AgentResult`	Outcome of a single agent invocation.
`flyte.ai.agents.AgentTool`	A normalized tool descriptor used by `flyte.ai.agents.Agent`.
`flyte.ai.agents.ConcurrencyError`	Raised when an `expected_sha` precondition does not match the current state.
`flyte.ai.agents.LLMMessage`	Provider-agnostic shape returned by `flyte.ai.agents.LLMCallable`.
`flyte.ai.agents.MCPServerSpec`	Declarative spec for a remote MCP server that exposes tools.
`flyte.ai.agents.MemoryMeta`	Per-file metadata sidecar (sha256, actor, timestamp, …) for a memory entry.
`flyte.ai.agents.MemoryStore`	Conversation transcript + path-addressed artifact memory backed by `flyte.io.Dir`.
`flyte.ai.agents.MemoryStoreError`	Base class for `flyte.ai.agents.MemoryStore` errors.
`flyte.ai.agents.ToolFn`	The tool under invocation, handed to a `flyte.ai.agents.ToolCallHandler`.
`flyte.ai.agents.agent.Agent`	A flyte-native tool-use agent harness.
`flyte.ai.agents.agent.AgentEvent`	Lightweight event emitted by the agent loop.
`flyte.ai.agents.memory.AccessDenied`	Raised when a write targets a read-only or reserved prefix.
`flyte.ai.agents.memory.ConcurrencyError`	Raised when an `expected_sha` precondition does not match the current state.
`flyte.ai.agents.memory.MemoryMeta`	Per-file metadata sidecar (sha256, actor, timestamp, …) for a memory entry.
`flyte.ai.agents.memory.MemoryStore`	Conversation transcript + path-addressed artifact memory backed by `flyte.io.Dir`.
`flyte.ai.agents.memory.MemoryStoreError`	Base class for `flyte.ai.agents.MemoryStore` errors.
`flyte.ai.agents.protocol.AgentResult`	Outcome of a single agent invocation.
`flyte.ai.chat.AgentChatAppEnvironment`	An `flyte.app.AppEnvironment` that spins up a FastAPI chat interface backed by any object satisfying the `flyte.ai.agents.AgentProtocol`.
`flyte.ai.chat.CustomTheme`	Declarative color theme for the Agent Chat UI.
`flyte.ai.chat.app.AgentChatAppEnvironment`	An `flyte.app.AppEnvironment` that spins up a FastAPI chat interface backed by any object satisfying the `flyte.ai.agents.AgentProtocol`.
`flyte.ai.chat.app.CustomTheme`	Declarative color theme for the Agent Chat UI.
`flyte.ai.mcp.FlyteMCPAppEnvironment`	Serve a Flyte-facing MCP server over HTTP (FastMCP + Starlette + Uvicorn).
`flyte.ai.mcp.MCPAppEnvironment`	Serve a FastMCP server over HTTP (Starlette + Uvicorn) or over stdio.
`flyte.app.AppEndpoint`	Embed an upstream app’s endpoint as an app parameter.
`flyte.app.AppEnvironment`	Configure a long-running app environment for APIs, dashboards, or model servers.
`flyte.app.ArtifactValue`	Use a published artifact as an app parameter value.
`flyte.app.ConnectorEnvironment`	Configure a connector environment for custom Flyte connectors.
`flyte.app.DeployedAppEnvironment`
`flyte.app.Domain`	Subdomain to use for the domain.
`flyte.app.Link`	Custom links to add to the app.
`flyte.app.Parameter`	Parameter for application.
`flyte.app.Port`
`flyte.app.RunOutput`	Use a run’s output for app parameters.
`flyte.app.Scaling`	Controls replica count and autoscaling behavior for app environments.
`flyte.app.Subdomain`	A subdomain that is resolved at deploy time, when the deployment project and domain are known.
`flyte.app.Timeouts`	Timeout configuration for the application.
`flyte.app.extras.FastAPIAppEnvironment`
`flyte.app.extras.FastAPIPassthroughAuthMiddleware`	FastAPI middleware that automatically sets Flyte auth metadata from request headers.
`flyte.app.extras.FlyteWebhookAppEnvironment`	A pre-built FastAPI app environment for common Flyte webhook operations.
`flyte.artifacts.Card`
`flyte.artifacts.Metadata`	Structured metadata for Flyte artifacts.
`flyte.clustered.ClusterFailurePolicy`	Failure and restart policy for the JobSet as a whole.
`flyte.clustered.ClusteredTaskEnvironment`	A TaskEnvironment that emits a Kubernetes JobSet for distributed multi-node training.
`flyte.clustered.ClusteredTaskTemplate`	Task template for `ClusteredTaskEnvironment`.
`flyte.clustered.JaxRun`	JAX multi-process runtime for a ClusteredTaskEnvironment.
`flyte.clustered.TorchRun`	TorchRun launcher configuration for a ClusteredTaskEnvironment.
`flyte.config.Config`	This the parent configuration object and holds all the underlying configuration object types.
`flyte.connectors.AsyncConnector`	This is the base class for all async connectors, and it defines the interface that all connectors must implement.
`flyte.connectors.AsyncConnectorExecutorMixin`	This mixin class is used to run the connector task locally, and it’s only used for local execution.
`flyte.connectors.ConnectorRegistry`	This is the registry for all connectors.
`flyte.connectors.ConnectorService`
`flyte.connectors.Resource`	This is the output resource of the job.
`flyte.connectors.ResourceMeta`	This is the metadata for the job.
`flyte.errors.ActionAbortedError`	This error is raised when an action was aborted, externally.
`flyte.errors.ActionNotFoundError`	This error is raised when the user tries to access an action that does not exist.
`flyte.errors.BaseRuntimeError`	Base class for all Union runtime errors.
`flyte.errors.CodeBundleError`	This error is raised when the code bundle cannot be created, for example when no files are found to bundle.
`flyte.errors.ConditionAlreadyExistsError`	This error is raised when the user tries to create a condition that already exists within the action.
`flyte.errors.ConditionFailedError`	This error is raised when a condition fails during execution.
`flyte.errors.ConditionNotFoundError`	This error is raised when the user tries to access a condition that does not exist.
`flyte.errors.ConditionTimedoutError`	This error is raised when a condition is not signaled within its specified timeout.
`flyte.errors.CustomError`	This error is raised when the user raises a custom error.
`flyte.errors.DeploymentError`	This error is raised when the deployment of a task fails, or some preconditions for deployment are not met.
`flyte.errors.GPUFaultError`	This error is raised when the backend attributed the task failure to a GPU or NVSwitch fault that the GPU health daemon observed on the node, such as an Xid 31 (a GPU memory page fault) or an Xid 79 (the GPU fell off the bus).
`flyte.errors.GPUFaultSystemError`	This error is raised when the GPU fault the backend attributed the failure to condemned the device or the node, for example an uncorrectable ECC error or a GPU that fell off the bus.
`flyte.errors.GPUFaultUserError`	This error is raised when the GPU fault the backend attributed the failure to was the workload’s own doing, for example an out-of-bounds access that the driver reported as an Xid 31.
`flyte.errors.ImageBuildError`	This error is raised when the image build fails.
`flyte.errors.ImagePullBackOffError`	This error is raised when the image cannot be pulled.
`flyte.errors.InitializationError`	This error is raised when the Union system is tried to access without being initialized.
`flyte.errors.InlineIOMaxBytesBreached`	This error is raised when the inline IO max bytes limit is breached.
`flyte.errors.InvalidImageNameError`	This error is raised when the image name is invalid.
`flyte.errors.InvalidPackageError`	Raised when an invalid system package is detected during image build.
`flyte.errors.LogsNotYetAvailableError`	This error is raised when the logs are not yet available for a task.
`flyte.errors.ModuleLoadError`	This error is raised when the module cannot be loaded, either because it does not exist or because of a syntax error.
`flyte.errors.NonRecoverableError`	Raised when an error is encountered that is not recoverable.
`flyte.errors.NotInTaskContextError`	This error is raised when the user tries to access the task context outside of a task.
`flyte.errors.OOMError`	This error is raised when the underlying task execution fails because of an out-of-memory error.
`flyte.errors.OnlyAsyncIOSupportedError`	This error is raised when the user tries to use sync IO in an async task.
`flyte.errors.ParameterMaterializationError`	This error is raised when the user tries to use a Parameter in an App, that has delayed Materialization, but the materialization fails.
`flyte.errors.PrimaryContainerNotFoundError`	This error is raised when the primary container is not found.
`flyte.errors.RemoteTaskNotFoundError`	This error is raised when the user tries to access a task that does not exist.
`flyte.errors.RemoteTaskUsageError`	This error is raised when the user tries to access a task that does not exist.
`flyte.errors.ResourceExhaustedError`
`flyte.errors.RestrictedTypeError`	This error is raised when the user uses a restricted type, for example current a Tuple is not supported for one value.
`flyte.errors.RetriesExhaustedError`	This error is raised when the underlying task execution fails after all retries have been exhausted.
`flyte.errors.RuntimeDataValidationError`	This error is raised when the user tries to access a resource that does not exist or is invalid.
`flyte.errors.RuntimeSystemError`	This error is raised when the underlying task execution fails because of a system error.
`flyte.errors.RuntimeUnknownError`	This error is raised when the underlying task execution fails because of an unknown error.
`flyte.errors.RuntimeUserError`	This error is raised when the underlying task execution fails because of an error in the user’s code.
`flyte.errors.SlowDownError`	This error is raised when the user tries to access a resource that does not exist or is invalid.
`flyte.errors.SyncTaskCallInAsyncContextError`	This error is raised when a sync task is invoked in a blocking way (`task(...)`) from inside an async task.
`flyte.errors.TaskInterruptedError`	This error is raised when the underlying task execution is interrupted.
`flyte.errors.TaskTimeoutError`	This error is raised when the underlying task execution runs for longer than the specified timeout.
`flyte.errors.TraceDoesNotAllowNestedTasksError`	This error is raised when the user tries to use a task from within a trace.
`flyte.errors.UnionRpcError`	This error is raised when communication with the Union server fails.
`flyte.extend.AsyncFunctionTaskTemplate`	A task template that wraps an asynchronous functions.
`flyte.extend.ImageBuildEngine`	ImageBuildEngine contains a list of builders that can be used to build an ImageSpec.
`flyte.extend.TaskTemplate`	Task template is a template for a task that can be executed.
`flyte.extras.BatchStats`	Monitoring statistics exposed by `DynamicBatcher.stats`.
`flyte.extras.ContainerTask`	This is an intermediate class that represents Flyte Tasks that run a container at execution time.
`flyte.extras.DynamicBatcher`	Batches records from many concurrent producers and runs them through a single async processing function, maximizing resource utilization.
`flyte.extras.Prompt`	Simple prompt record with built-in token estimation.
`flyte.extras.Sleep`	Route a task to the backend `core-sleep` plugin.
`flyte.extras.SleepTask`
`flyte.extras.TokenBatcher`	Token-aware batcher for LLM inference workloads.
`flyte.extras.shell.FlagSpec`	How to render a typed input as a CLI flag in `{flags.<name>}`.
`flyte.extras.shell.Glob`	A multi-file output bundle.
`flyte.extras.shell.Stderr`	Capture the task’s stderr as a typed output.
`flyte.extras.shell.Stdout`	Capture the task’s stdout as a typed output.
`flyte.extras.webhooks.EventType`	Base for event constants: a real `str`, usable anywhere a pattern is.
`flyte.extras.webhooks.Provider`	Everything core needs to accept one product’s webhooks.
`flyte.extras.webhooks.RunOnceResult`	The run covering a dedupe key, and whether this call created it.
`flyte.extras.webhooks.SignatureError`	Raised when an inbound delivery fails verification or cannot be parsed.
`flyte.extras.webhooks.WebhookAppEnvironment`	Dashboard plus a verified webhook receiver for one or more providers.
`flyte.extras.webhooks.WebhookEvent`	One inbound webhook, normalized across providers.
`flyte.extras.webhooks.WebhookPluginError`	Base class for all errors raised by this plugin.
`flyte.git.GitStatus`	A class representing the status of a git repository.
`flyte.io.DataFrame`	A Flyte meta DataFrame object, that wraps all other dataframe types (usually available as plugins, pandas.DataFrame and pyarrow.Table are supported natively, just install these libraries).
`flyte.io.Dir`	A generic directory class representing a directory with files of a specified format.
`flyte.io.EmptyDir`	A sentinel `flyte.io.Dir` representing `no directory was produced`.
`flyte.io.File`	A generic file class representing a file with a specified format.
`flyte.io.HashFunction`	A hash method that wraps a user-provided function to compute hashes.
`flyte.io.extend.DataFrameDecoder`
`flyte.io.extend.DataFrameEncoder`
`flyte.io.extend.DataFrameTransformerEngine`	Think of this transformer as a higher-level meta transformer that is used for all the dataframe types.
`flyte.models.ActionID`	A class representing the ID of an Action, nested within a Run.
`flyte.models.ActionPhase`	Represents the execution phase of a Flyte action (run).
`flyte.models.CheckpointPaths`	Paths the platform provides for this task’s checkpoint output and optional previous-attempt input.
`flyte.models.CodeBundle`	A class representing a code bundle for a task.
`flyte.models.GroupData`
`flyte.models.NativeInterface`	A class representing the native interface for a task.
`flyte.models.PathRewrite`	Configuration for rewriting paths during input loading.
`flyte.models.RawDataPath`	A class representing the raw data path for a task.
`flyte.models.SerializationContext`	This object holds serialization time contextual information, that can be used when serializing the task and various parameters of a tasktemplate.
`flyte.models.TaskContext`	A context class to hold the current task executions context.
`flyte.notify.Email`	Send email notifications.
`flyte.notify.NamedDelivery`	Use a pre-configured delivery channel by name.
`flyte.notify.NamedRule`	Reference a pre-defined notification rule by name.
`flyte.notify.Notification`	Base notification class.
`flyte.notify.Slack`	Send Slack notifications with optional Block Kit formatting.
`flyte.notify.Teams`	Send Microsoft Teams notifications with optional Adaptive Cards.
`flyte.notify.Webhook`	Send custom HTTP webhook notifications (most flexible option).
`flyte.prefetch.HuggingFaceModelInfo`	Information about a HuggingFace model to store.
`flyte.prefetch.ShardConfig`	Configuration for model sharding.
`flyte.prefetch.StoredModelInfo`	Information about a stored model.
`flyte.prefetch.VLLMShardArgs`	Arguments for sharding a model using vLLM.
`flyte.remote.Action`	A class representing an action.
`flyte.remote.ActionDetails`	A class representing an action.
`flyte.remote.ActionInputs`	A class representing the inputs of an action.
`flyte.remote.ActionOutputs`	A class representing the outputs of an action.
`flyte.remote.App`
`flyte.remote.Artifact`	A published artifact in the Flyte artifact service: a typed value (stored as a Flyte literal) addressed by org/project/domain/name/version.
`flyte.remote.Condition`	A remote Condition registered within an action of a run.
`flyte.remote.Project`	A class representing a project in the Union API.
`flyte.remote.Run`	A class representing a run of a task.
`flyte.remote.RunDetails`	A class representing a run of a task.
`flyte.remote.Secret`
`flyte.remote.Settings`	Hierarchical configuration settings with inheritance support.
`flyte.remote.Task`
`flyte.remote.TaskDetails`
`flyte.remote.TimeFilter`	Filter for time-based fields (e.g. created_at, updated_at).
`flyte.remote.Trigger`	Represents a trigger in the Flyte platform.
`flyte.remote.TriggerDetails`
`flyte.remote.User`	Represents a user in the Flyte platform.
`flyte.report.Report`
`flyte.report.Timeline`	Append a best-effort chronological timeline to a tab of the task report.
`flyte.sandbox.CodeTaskTemplate`	A sandboxed task created from a code string rather than a decorated function.
`flyte.sandbox.ImageConfig`	Configuration for Docker image building at runtime.
`flyte.sandbox.SandboxedConfig`	Configuration for a sandboxed task executed via Monty.
`flyte.sandbox.SandboxedTaskTemplate`	A task template that executes the function body in a Monty sandbox.
`flyte.storage.ABFS`	Any Azure Blob Storage specific configuration.
`flyte.storage.GCS`	Any GCS specific configuration.
`flyte.storage.S3`	S3 specific configuration.
`flyte.storage.Storage`	Data storage configuration that applies across any provider.
`flyte.syncify.Syncify`	A decorator to convert asynchronous functions or methods into synchronous ones.
`flyte.types.FlytePickle`	This type is only used by flytekit internally.
`flyte.types.TypeEngine`	Core Extensible TypeEngine of Flytekit.
`flyte.types.TypeTransformer`	Base transformer type that should be implemented for every python native type that can be handled by flytekit.
`flyte.types.TypeTransformerFailedError`

Protocols

Protocol	Description
`flyte.AppHandle`	Protocol defining the common interface between local and remote app handles.
`flyte.CachePolicy`	Protocol for custom cache version strategies.
`flyte.Link`
`flyte.ai.agents.AgentProtocol`	Minimal protocol that any agent must satisfy to work with `flyte.ai.chat.AgentChatAppEnvironment`.
`flyte.ai.agents.protocol.AgentProtocol`	Minimal protocol that any agent must satisfy to work with `flyte.ai.chat.AgentChatAppEnvironment`.
`flyte.artifacts.Artifact`	Protocol for objects wrapped with Flyte metadata.
`flyte.extend.ImageBuilder`
`flyte.extend.ImageChecker`
`flyte.extras.CostEstimator`	Protocol for records that can estimate their own processing cost.
`flyte.extras.TokenEstimator`	Protocol for records that can estimate their own token count.
`flyte.extras.webhooks.testing.ProviderFactory`	What a plugin’s exported provider class must look like.
`flyte.types.Renderable`

Functions

Function	Description
`flyte.AMD_GPU()`	Create an AMD GPU device instance.
`flyte.GPU()`	Create a GPU device instance.
`flyte.HABANA_GAUDI()`	Create a Habana Gaudi device instance.
`flyte.Neuron()`	Create a Neuron device instance.
`flyte.TPU()`	Create a TPU device instance.
`flyte.build()`	Build an image.
`flyte.build_images()`	Build the images for the given environment(s).
`flyte.ctx()`	Returns the current flyte.models.TaskContext when running inside a task.
`flyte.current_domain()`	Returns the current domain from Runtime environment (on the cluster) or from the initialized configuration.
`flyte.current_project()`	Returns the current project from the Runtime environment (on the cluster) or from the initialized configuration.
`flyte.custom_context()`	Synchronous context manager to set input context for tasks spawned within this block.
`flyte.deploy()`	Deploy the given environment or list of environments.
`flyte.get_custom_context()`	Get the current input context.
`flyte.group()`	Create a new group with the given name.
`flyte.init()`	Initialize the Flyte system with the given configuration.
`flyte.init_from_api_key()`	Initialize the Flyte system using an API key for authentication.
`flyte.init_from_config()`	Initialize the Flyte system using a configuration file or Config object.
`flyte.init_in_cluster()`	Initialize the Flyte system from inside a task pod, and return the kwargs used to build the controller that enqueues and watches child actions.
`flyte.init_passthrough()`	Initialize the Flyte system with passthrough authentication.
`flyte.is_control_plane_available()`	True when this process can submit work to a Flyte control plane — `flyte.run` launches real remote runs whose actions can be inspected, awaited, and replayed (recovered/forked).
`flyte.latest_checkpoint()`	Return the file under root matching glob_pattern with the largest `key(path)`, or `None`.
`flyte.load_interactive_ctx()`	Restore the task execution context from the config file written by a debug-mode task pod.
`flyte.load_plugin_config()`	Load a plugin config instance from a YAML file.
`flyte.map()`	Map a function over the provided arguments with concurrent execution.
`flyte.new_condition()`	Create a condition that can be awaited in a workflow.
`flyte.rerun()`	Re-run a prior run, returning a new `Run`.
`flyte.run()`	Run a task with the given parameters, or fire a deployed trigger on demand.
`flyte.run_python_script()`	Package and run a Python script on a remote Flyte cluster.
`flyte.serve()`	Serve a Flyte app using an AppEnvironment.
`flyte.trace()`	A decorator that traces function execution with timing information.
`flyte.version()`	Returns the version of the Flyte SDK.
`flyte.with_runcontext()`	Launch a new run with the given parameters as the context.
`flyte.with_servecontext()`	Create a serve context with custom configuration.
`flyte.ai.agents.tool()`	Wrap a task, `@flyte.trace` helper, plain callable, or `LazyEntity` as a `flyte.ai.agents.AgentTool`.
`flyte.ai.chat.build_chat_html()`	Build the full chat HTML with the given title and optional custom_css.
`flyte.ai.mcp.resolve_tools()`	Return the set of MCP tool names to expose.
`flyte.app.ctx()`	Returns the current app context.
`flyte.app.get_parameter()`	Get parameters for application or endpoint.
`flyte.artifacts.new()`	Wrap an object with Flyte metadata while preserving its type interface.
`flyte.clustered.jax_initialize()`	Initialize `jax.distributed` for this clustered task’s process topology.
`flyte.config.auto()`	Automatically constructs the Config Object.
`flyte.config.set_if_exists()`	Given a dict `d` sets the key `k` with value of config `v`, if the config value `v` is set and return the updated dictionary.
`flyte.connectors.utils.convert_to_flyte_phase()`	Convert the state from the connector to the phase in flyte.
`flyte.connectors.utils.is_terminal_phase()`	Return true if the phase is terminal.
`flyte.connectors.utils.print_metadata()`
`flyte.durable.now()`	Returns the current time for every unique invocation of durable_time.
`flyte.durable.sleep()`	durable_sleep enables the process to sleep for `seconds` seconds even if the process recovers from a crash.
`flyte.durable.time()`	Returns the current time for every unique invocation of durable_time.
`flyte.errors.silence_polling_error()`	Suppress specific polling errors in the event loop.
`flyte.extend.download_code_bundle()`	Downloads the code bundle if it is not already downloaded.
`flyte.extend.get_proto_extended_resources()`	TODO Implement partitioning logic string handling for GPU.
`flyte.extend.get_proto_resources()`	Get main resources IDL representation from the resources object.
`flyte.extend.is_initialized()`	Check if the system has been initialized.
`flyte.extend.lazy_module()`	This function is used to lazily import modules.
`flyte.extend.pod_spec_from_resources()`
`flyte.extras.serialize()`	Translate a single task to its wire TaskSpec, offline and code-agnostic.
`flyte.extras.serialize_env()`	Serialize every task in an environment.
`flyte.extras.shell.create()`	Wrap a CLI tool packaged in a container as a Flyte task.
`flyte.extras.webhooks.blocking_run()`	Return the run that blocks this key, or None.
`flyte.extras.webhooks.constant_time_equals()`	Compare two credentials in constant time, without raising.
`flyte.extras.webhooks.hex_hmac_sha256()`	Hex HMAC-SHA256 of the raw body — the scheme most products use.
`flyte.extras.webhooks.json_body()`	Parse a JSON body into a dict, raising `SignatureError` when it is not one.
`flyte.extras.webhooks.lower_headers()`	Lowercase header keys, since HTTP header names are case-insensitive.
`flyte.extras.webhooks.run_once()`	Launch `task` once for `key`, returning the run that covers it.
`flyte.extras.webhooks.testing.assert_provider_conforms()`	Assert that a provider plugin implements the common webhook contract.
`flyte.git.config_from_root()`	Get the config file from the git root directory.
`flyte.models.generate_random_name()`	Generate a random name for the task.
`flyte.models.is_clustered_worker_env()`	True for any worker process of a clustered/jobset task, regardless of which launcher started it.
`flyte.prefetch.hf_model()`	Store a HuggingFace model to remote storage.
`flyte.remote.auth_metadata()`	This context manager allows you to pass contextualized auth metadata downstream to the Flyte authentication system.
`flyte.remote.upload_dir()`	Uploads a directory to a remote location and returns the remote URI.
`flyte.remote.upload_file()`	Uploads a file to a remote location and returns the remote URI.
`flyte.report.abbreviate()`	HTML-escape `value` for a report row.
`flyte.report.current_report()`	Get the current report.
`flyte.report.duration_ms()`	Format the gap between two ISO-8601 timestamps as `"<n> ms"` (best-effort).
`flyte.report.flush()`	Flush the report.
`flyte.report.get_tab()`	Get a tab by name.
`flyte.report.log()`	Log content to the main tab.
`flyte.report.replace()`	Get the report.
`flyte.sandbox.create()`	Create a stateless Python code sandbox.
`flyte.sandbox.orchestrate_local()`	One-shot local execution of a code string in the Monty sandbox.
`flyte.sandbox.orchestrator_from_str()`	Create a reusable sandboxed task from a code string.
`flyte.storage.exists()`	Check if a path exists.
`flyte.storage.exists_sync()`
`flyte.storage.get()`
`flyte.storage.get_configured_fsspec_kwargs()`
`flyte.storage.get_random_local_directory()`	pathlib.Path: a random directory.
`flyte.storage.get_random_local_path()`	Use file_path_or_file_name, when you want a random directory, but want to preserve the leaf file name.
`flyte.storage.get_stream()`	Get a stream of data from a remote location.
`flyte.storage.get_underlying_filesystem()`
`flyte.storage.is_remote()`	Let’s find a replacement.
`flyte.storage.join()`	Join multiple paths together.
`flyte.storage.open()`	Asynchronously open a file and return an async context manager.
`flyte.storage.put()`
`flyte.storage.put_stream()`	Put a stream of data to a remote location.
`flyte.types.guess_interface()`	Returns the interface of the task with guessed types, as types may not be present in current env.
`flyte.types.literal_string_repr()`	This method is used to convert a literal map to a string representation.

Packages

Package	Description
`flyte`	Flyte SDK for authoring compound AI applications, services and workflows.
`flyte.ai.agents`	flyte.ai.agents — Agent abstractions for Flyte apps.
`flyte.ai.agents.agent`	Agent — a flyte-native tool-use agent harness.
`flyte.ai.agents.memory`	Dir-backed memory for `flyte.ai.agents.Agent`.
`flyte.ai.agents.protocol`	Agent protocol for the flyte.ai.agents module.
`flyte.ai.chat`	flyte.ai.chat — FastAPI chat UI and HTML/CSS assets for Flyte agents.
`flyte.ai.chat.app`	AgentChatAppEnvironment — FastAPI-based chat UI for any Agent.
`flyte.ai.mcp`
`flyte.app`
`flyte.app.extras`
`flyte.artifacts`	Artifacts module.
`flyte.clustered`
`flyte.config`
`flyte.connectors`
`flyte.connectors.utils`
`flyte.durable`	Flyte durable utilities.
`flyte.errors`	Exceptions raised by Union.
`flyte.extend`
`flyte.extras`	Flyte extras package.
`flyte.extras.shell`	Shell task — wrap a CLI tool packaged in a container image.
`flyte.extras.webhooks`	Receive SaaS webhooks in Flyte, and turn them into runs.
`flyte.extras.webhooks.testing`	Conformance harness — enforce the common provider format.
`flyte.git`
`flyte.io`	This package contains additional data types beyond the primitive data types in python to abstract data flow of large datasets in Union.
`flyte.io.extend`
`flyte.models`
`flyte.notify`	Task Notifications API for Flyte 2.0.
`flyte.prefetch`	Prefetch utilities for Flyte.
`flyte.remote`	Remote Entities that are accessible from the Union Server once deployed or created.
`flyte.report`
`flyte.sandbox`	Sandbox utilities for running isolated code inside Flyte tasks.
`flyte.storage`
`flyte.syncify`	This module provides the `syncify` decorator and the `Syncify` class.
`flyte.types`	The Flyte type system provides a way to define, transform, and manipulate types in Flyte workflows.