API Documentation

Use the Python SDK when you want Tamarind to validate, package, upload, build, and monitor a Custom Tool for you. The SDK keeps transport details such as upload URLs and concurrency headers out of application code.

Install and connect

1pip install --upgrade tamarind-cli
2export TAMARIND_API_KEY="YOUR_API_KEY"

1from tamarind import Tamarind
2
3# TAMARIND_API_KEY is read automatically.
4with Tamarind(api_base="https://app.tamarind.bio/api/") as client:
5    tools = client.custom_tools.list()
6    print([tool.name for tool in tools.items])

Method index

Jump to a client, tool, or version method.

Client and collection

Create a configured SDK session, then create, retrieve, or list organization-visible tools.

Tamarind

Tamarind(
    *,
    api_key: str | None = None,
    api_base: str | None = None,
    profile: str | None = None,
    timeout: float = 120.0,
)

Creates an SDK session. Use it as a context manager so network resources are closed automatically.

Parameters

api_key str | None, optional: API key. When omitted, the active Tamarind profile or TAMARIND_API_KEY is used.
api_base str | None, optional: Overrides the configured organization API base URL. Include the trailing /api/ path; omit it to use the host from the selected profile.
profile str | None, optional: Loads credentials and host settings from this named Tamarind CLI profile. Omit it to use the active profile.
timeout float, optional: Maximum time in seconds for each API request before it fails. Defaults to 120 seconds.

Returns

Tamarind: A configured client. Use custom_tools to create and retrieve tools; leaving the context manager closes its network resources.

Raises

AuthError
APIError

Examples

1with Tamarind() as client:
2    tool = client.custom_tools.get("my-custom-tool")

client.custom_tools.create

client.custom_tools.create(
    name: str,
    *,
    display_name: str = ...,
    description: str = ...,
    gpu_type: str = ...,
    memory: str = ...,
    cpu: int = ...,
) -> CustomTool

Creates a draft Custom Tool. The name becomes the stable value used as type when submitting jobs.

Parameters

name str: Unique tool name.
display_name str, optional: Human-readable name shown in the UI and tool listings. Omit it to use the server default.
description str, optional: Short explanation shown to people deciding when to use the tool. Omit it to use the server default.
gpu_type str, optional: GPU allocated to each run: None, T4, L4, L40S, A10, or A100. Omit it to use the server default.
memory str, optional: Memory allocated to each run, such as 8Gi, 32Gi, 96Gi, or 180Gi. Omit it to use the server default.
cpu int, optional: Number of CPU cores allocated to each run. Omit it to use the server default.

Returns

CustomTool: The new draft tool, including its stable name, metadata, current status, and resource settings.

View fields

name str: Stable tool name used as type when submitting jobs.
display_name str: Human-readable name shown in the UI.
description str: Explanation of what the tool does.
functions tuple[str, ...]: Short capability labels used for discovery.
status CustomToolStatus: Current lifecycle state: Draft, Building, or Deployed.
gpu_type str: GPU allocated to each run, or None for CPU-only execution.
memory str: Memory allocated to each run, expressed in GiB.
cpu int: Number of CPU cores allocated to each run.
home_disk_gi int: Writable home-disk capacity in GiB.
max_runtime_seconds int | None: Maximum runtime for a job, or None when no tool-specific limit is set.
has_source bool: Whether source has been uploaded for the current draft.
published bool: Whether the tool has a published default version.
auto_publish bool: Whether successful builds become the default automatically.
est_time str: Estimated runtime in H:M:S format.
paper_url str: Related publication URL, when configured.
tags tuple[str, ...]: Search and discovery tags.
default_version str | None: Published version name used when a job omits version.
created_at str: ISO 8601 creation timestamp.
updated_at str: ISO 8601 timestamp of the latest update.
can_edit bool: Whether the current credentials may change this tool.
can_build bool: Whether the current credentials may build this tool.
version Version | None: Requested or latest version, or None when the tool has never been built.

Raises

CustomToolExistsError
ValidationError
AuthError
APIError

Examples

1tool = client.custom_tools.create(
2    "my-custom-tool",
3    display_name="My Custom Tool",
4)

client.custom_tools.get

client.custom_tools.get(
    name: str,
    *,
    version: str | None = None,
) -> CustomTool

Gets one tool visible to the current organization, including its latest version by default or a selected version such as v2.

Parameters

name str: Tool name returned by create() or list().
version str | None, optional: Public version label such as v2. Omit it to include the latest version.

Returns

CustomTool: The tool snapshot, including metadata, status, published default_version, and the requested or latest Version when one exists.

View fields

name str: Stable tool name used as type when submitting jobs.
display_name str: Human-readable name shown in the UI.
description str: Explanation of what the tool does.
functions tuple[str, ...]: Short capability labels used for discovery.
status CustomToolStatus: Current lifecycle state: Draft, Building, or Deployed.
gpu_type str: GPU allocated to each run, or None for CPU-only execution.
memory str: Memory allocated to each run, expressed in GiB.
cpu int: Number of CPU cores allocated to each run.
home_disk_gi int: Writable home-disk capacity in GiB.
max_runtime_seconds int | None: Maximum runtime for a job, or None when no tool-specific limit is set.
has_source bool: Whether source has been uploaded for the current draft.
published bool: Whether the tool has a published default version.
auto_publish bool: Whether successful builds become the default automatically.
est_time str: Estimated runtime in H:M:S format.
paper_url str: Related publication URL, when configured.
tags tuple[str, ...]: Search and discovery tags.
default_version str | None: Published version name used when a job omits version.
created_at str: ISO 8601 creation timestamp.
updated_at str: ISO 8601 timestamp of the latest update.
can_edit bool: Whether the current credentials may change this tool.
can_build bool: Whether the current credentials may build this tool.
version Version | None: Requested or latest version, or None when the tool has never been built.

Raises

CustomToolNotFoundError
AuthError
APIError

Examples

1tool = client.custom_tools.get("my-custom-tool", version="v2")

client.custom_tools.list

client.custom_tools.list(
    *,
    status: str | None = None,
    published: bool | None = None,
    limit: int = 50,
    cursor: str | None = None,
) -> Page[CustomTool]

Lists tools visible to the current organization. Use next_cursor to fetch the next page.

Parameters

status str | None, optional: Returns only tools in the selected Draft, Building, or Deployed state. Omit it to include every state.
published bool | None, optional: Use True for published tools or False for unpublished tools. Omit it to include both.
limit int, optional: Maximum number of tools returned in one page. Defaults to 50.
cursor str | None, optional: Continuation cursor from the previous page. Omit it when requesting the first page.

Returns

Page[CustomTool]: items contains the tools in this page. Pass next_cursor to the next list() call; it is None after the final page.

View fields

items tuple[T, ...]: Resources returned in the current page.
next_cursor str | None: Pass this value to the next list call. None means there are no more results.

Raises

AuthError
APIError

Examples

1page = client.custom_tools.list(published=True)
2for tool in page.items:
3    print(tool.name, tool.default_version)

CustomTool

Inspect and change one tool, validate source locally, build versions, and submit test jobs.

tool.refresh

tool.refresh() -> CustomTool

Fetches a fresh snapshot of the same tool. The SDK rejects the refresh if that name was deleted and recreated.

Returns

CustomTool: A fresh snapshot containing the tool's current metadata, status, resource settings, and default_version.

View fields

name str: Stable tool name used as type when submitting jobs.
display_name str: Human-readable name shown in the UI.
description str: Explanation of what the tool does.
functions tuple[str, ...]: Short capability labels used for discovery.
status CustomToolStatus: Current lifecycle state: Draft, Building, or Deployed.
gpu_type str: GPU allocated to each run, or None for CPU-only execution.
memory str: Memory allocated to each run, expressed in GiB.
cpu int: Number of CPU cores allocated to each run.
home_disk_gi int: Writable home-disk capacity in GiB.
max_runtime_seconds int | None: Maximum runtime for a job, or None when no tool-specific limit is set.
has_source bool: Whether source has been uploaded for the current draft.
published bool: Whether the tool has a published default version.
auto_publish bool: Whether successful builds become the default automatically.
est_time str: Estimated runtime in H:M:S format.
paper_url str: Related publication URL, when configured.
tags tuple[str, ...]: Search and discovery tags.
default_version str | None: Published version name used when a job omits version.
created_at str: ISO 8601 creation timestamp.
updated_at str: ISO 8601 timestamp of the latest update.
can_edit bool: Whether the current credentials may change this tool.
can_build bool: Whether the current credentials may build this tool.
version Version | None: Requested or latest version, or None when the tool has never been built.

Raises

CustomToolNotFoundError
StaleCustomToolError
APIError

Examples

1tool = tool.refresh()

tool.update

tool.update(
    *,
    display_name=...,
    description=...,
    functions=...,
    gpu_type=...,
    memory=...,
    cpu=...,
    home_disk_gi=...,
    auto_publish=...,
    est_time=...,
    paper_url=...,
    tags=...,
) -> CustomTool

Updates the supplied metadata and resource fields. Only concrete values are applied; omitted arguments and None leave the current value unchanged.

Parameters

display_name str | None, optional: Human-readable name. Omitting this argument or passing None leaves it unchanged. To clear it, pass an empty string ("").
description str | None, optional: Tool description. Omitting this argument or passing None leaves it unchanged. To clear it, pass an empty string ("").
functions list[str] | None, optional: Short capability labels. Omitting this argument or passing None leaves it unchanged. To clear it, pass an empty list ([]).
gpu_type str | None, optional: GPU type: "None", T4, L4, L40S, A10, or A100. Use the string "None" for no GPU. Omitting this argument or passing None leaves it unchanged.
memory str | None, optional: 8Gi, 12Gi, 24Gi, 32Gi, 48Gi, 64Gi, 90Gi, 96Gi, or 180Gi. Omitting this argument or passing None leaves it unchanged.
cpu int | None, optional: Requested CPU count. Omitting this argument or passing None leaves it unchanged.
home_disk_gi int | None, optional: Writable home-disk size in GiB. Omitting this argument or passing None leaves it unchanged.
auto_publish bool | None, optional: Whether successful builds become the default automatically. Pass False to disable it. Omitting this argument or passing None leaves it unchanged.
est_time str | None, optional: Estimated runtime in H:M:S format, for example 0:01:00. Omitting this argument or passing None leaves it unchanged. To clear it, pass an empty string ("").
paper_url str | None, optional: Related publication URL. Omitting this argument or passing None leaves it unchanged. To clear it, pass an empty string ("").
tags list[str] | None, optional: Search and discovery tags. Omitting this argument or passing None leaves it unchanged. To clear it, pass an empty list ([]).

Returns

CustomTool: The saved tool snapshot after applying the supplied changes. Omitted fields retain their previous values.

View fields

name str: Stable tool name used as type when submitting jobs.
display_name str: Human-readable name shown in the UI.
description str: Explanation of what the tool does.
functions tuple[str, ...]: Short capability labels used for discovery.
status CustomToolStatus: Current lifecycle state: Draft, Building, or Deployed.
gpu_type str: GPU allocated to each run, or None for CPU-only execution.
memory str: Memory allocated to each run, expressed in GiB.
cpu int: Number of CPU cores allocated to each run.
home_disk_gi int: Writable home-disk capacity in GiB.
max_runtime_seconds int | None: Maximum runtime for a job, or None when no tool-specific limit is set.
has_source bool: Whether source has been uploaded for the current draft.
published bool: Whether the tool has a published default version.
auto_publish bool: Whether successful builds become the default automatically.
est_time str: Estimated runtime in H:M:S format.
paper_url str: Related publication URL, when configured.
tags tuple[str, ...]: Search and discovery tags.
default_version str | None: Published version name used when a job omits version.
created_at str: ISO 8601 creation timestamp.
updated_at str: ISO 8601 timestamp of the latest update.
can_edit bool: Whether the current credentials may change this tool.
can_build bool: Whether the current credentials may build this tool.
version Version | None: Requested or latest version, or None when the tool has never been built.

Raises

ValidationError
StaleCustomToolError
APIError

Examples

1tool = tool.update(
2    description="Summarize a protein sequence",
3    est_time="0:01:00",
4    tags=["sequence", "summary"],
5)

tool.validate

tool.validate(folder: str | Path) -> ValidationReport

Checks the source folder locally without uploading or starting a build.

Parameters

folder str | Path: Folder containing Dockerfile, run.sh, and config.json at its root.

Returns

ValidationReport: valid is true when the folder can be built. errors contains blocking problems; warnings contains non-blocking guidance.

View fields

valid bool: True when no blocking validation errors were found.
errors tuple[ValidationProblem, ...]: Blocking problems. Each item contains code, path, and message.
warnings tuple[ValidationProblem, ...]: Non-blocking guidance. Each item contains code, path, and message.

Examples

1report = tool.validate("./my-custom-tool")
2if not report.valid:
3    for problem in report.errors:
4        print(problem.path, problem.message)

tool.build

tool.build(
    folder: str | Path,
    *,
    idempotency_key: str | None = None,
    source_timeout: float = 180.0,
) -> BuildResult

Validates and packages the folder, uploads it, and requests a version build. Upload URLs and integrity checks are handled internally.

Parameters

folder str | Path: Source folder to package and build.
idempotency_key str | None, optional: Stable key used to retry the same build request without creating a duplicate. Omit it for a normal one-time build.
source_timeout float, optional: Maximum time in seconds allowed for uploading the source archive. Defaults to 180 seconds.

Returns

BuildResult: action is build, reuse_image, or unchanged. version is the exact Version to monitor, test, or publish.

View fields

action BuildAction: build for a new image build, reuse_image for a new version using the current image, or unchanged when the source already has a version.
version Version: Exact version to monitor, test, or publish.

Raises

ValidationError
CustomToolUploadError
StaleCustomToolError
APIError

Examples

1build = tool.build("./my-custom-tool")
2version = build.version.monitor(timeout=1800)

tool.test

tool.test(
    settings: dict[str, Any],
    *,
    version: str | None = None,
    name: str | None = None,
) -> CustomToolTestJob

Submits a test job. Omit version to use the published default, or pass a version label such as v2 to run it before publishing.

Parameters

settings dict[str, Any]: Values matching the inputs declared in config.json.
version str | None, optional: Exact Version.version returned by build(), versions(), or get_version(). Omit it to use the published default.
name str | None, optional: Unique name for the test job. Omit it to let the SDK generate one.

Returns

CustomToolTestJob: An immediate submission receipt with id, job_name, job_type, status, and created_at. Use job_name with the normal job-status and result APIs.

View fields

id str: Job identifier.
job_name str: Job name used with status and result APIs.
job_type str: Custom Tool name submitted as the job type.
status str: Initial job status.
created_at str: ISO 8601 submission timestamp.

Raises

ValidationError
CustomToolNotDeployableError
APIError

Examples

1job = tool.test(
2    {"sequence": "MKTAYIAKQRQISFVKSHFSRQ"},
3    version=version.version,
4)
5print(job.job_name, job.status)

tool.versions

tool.versions(
    *,
    status: str | None = None,
    limit: int = 50,
    cursor: str | None = None,
) -> Page[Version]

Lists versions belonging to this tool.

Parameters

status str | None, optional: Returns only versions in the selected Queued, Running, Complete, or Stopped state. Omit it to include every state.
limit int, optional: Maximum number of versions returned in one page. Defaults to 50.
cursor str | None, optional: Continuation cursor from the previous page. Omit it when requesting the first page.

Returns

Page[Version]: items contains the versions in this page. Pass next_cursor to the next versions() call; it is None after the final page.

View fields

items tuple[T, ...]: Resources returned in the current page.
next_cursor str | None: Pass this value to the next list call. None means there are no more results.

Raises

StaleCustomToolError
APIError

Examples

1versions = tool.versions(status="Complete")
2for version in versions.items:
3    print(version.version, version.status)

tool.get_version

tool.get_version(version: str) -> Version

Gets one exact version using its public label, such as v2.

Parameters

version str: Version.version returned by build() or versions().

Returns

Version: The exact version identified by version, including its build status, timestamps, origin, and any terminal error.

View fields

version str: Public version label, such as v3; use it as version when submitting a job.
status VersionStatus: Build state: Queued, Running, Complete, or Stopped.
terminal bool: Whether the build has reached a final state and will not run again.
origin str: How this version was created, such as a source build or image reuse.
created_at str: ISO 8601 creation timestamp.
started_at str: ISO 8601 timestamp when build processing began.
completed_at str | None: ISO 8601 completion timestamp, or None while the build is active.
error BuildError | None: Structured failure code and message when the build stops unsuccessfully.
tool_name str: Name of the Custom Tool that owns this version.

Raises

ValidationError
CustomToolNotFoundError
APIError

Examples

1version = tool.get_version(build.version.version)

tool.delete

tool.delete() -> None

Deletes the tool and releases its name for reuse.

Returns

None: The call returns after the tool has been deleted successfully.

Raises

StaleCustomToolError
CustomToolBuildInProgressError
APIError

Examples

1tool.delete()

Version

Monitor and operate on the exact version returned by a build or version lookup.

version.refresh

version.refresh() -> Version

Fetches the latest status for this version without allowing a recreated tool with the same name to replace it.

Returns

Version: A fresh snapshot containing the version's current build status, timestamps, and any terminal error.

View fields

version str: Public version label, such as v3; use it as version when submitting a job.
status VersionStatus: Build state: Queued, Running, Complete, or Stopped.
terminal bool: Whether the build has reached a final state and will not run again.
origin str: How this version was created, such as a source build or image reuse.
created_at str: ISO 8601 creation timestamp.
started_at str: ISO 8601 timestamp when build processing began.
completed_at str | None: ISO 8601 completion timestamp, or None while the build is active.
error BuildError | None: Structured failure code and message when the build stops unsuccessfully.
tool_name str: Name of the Custom Tool that owns this version.

Raises

CustomToolNotFoundError
APIError

Examples

1version = version.refresh()

version.logs

version.logs(*, cursor: str | None = None) -> BuildLogPage

Reads one page of build events. Pass next_cursor to continue without rereading earlier pages.

Parameters

cursor str | None, optional: Continuation cursor from the previous log page. Omit it to read from the beginning.

Returns

BuildLogPage: items contains this page of build events. status and error describe the current build; next_cursor continues from the last event read.

View fields

items tuple[BuildEvent, ...]: Build events, each containing message and Unix-millisecond timestamp.
status VersionStatus: Current build state.
next_cursor str | None: Cursor for the next log page, or None when currently caught up.
error BuildError | None: Structured failure code and message when available.

Raises

CustomToolNotFoundError
APIError

Examples

1page = version.logs()
2for event in page.items:
3    print(event.message)

version.monitor

version.monitor(
    *,
    timeout: float | None,
    interval: float = 2.0,
    on_event: Callable[[BuildEvent], None] | None = None,
) -> Version

Polls until the build reaches Complete or Stopped. Use monitor_async() with the same arguments in asynchronous applications.

Parameters

timeout float | None: Maximum wait in seconds. None waits without a deadline.
interval float, optional: Seconds to wait between status checks. Defaults to 2 seconds.
on_event callable | None, optional: Function called once for each newly received build-log event. Omit it when no live log handling is needed.

Returns

Version: The completed version snapshot. Failed builds raise CustomToolBuildFailedError, and an expired deadline raises CustomToolBuildTimeoutError.

View fields

version str: Public version label, such as v3; use it as version when submitting a job.
status VersionStatus: Build state: Queued, Running, Complete, or Stopped.
terminal bool: Whether the build has reached a final state and will not run again.
origin str: How this version was created, such as a source build or image reuse.
created_at str: ISO 8601 creation timestamp.
started_at str: ISO 8601 timestamp when build processing began.
completed_at str | None: ISO 8601 completion timestamp, or None while the build is active.
error BuildError | None: Structured failure code and message when the build stops unsuccessfully.
tool_name str: Name of the Custom Tool that owns this version.

Raises

CustomToolBuildFailedError
CustomToolBuildTimeoutError
APIError

Examples

1version = build.version.monitor(
2    timeout=1800,
3    on_event=lambda event: print(event.message),
4)

version.monitor_async

await version.monitor_async(
    *,
    timeout: float | None,
    interval: float = 2.0,
    on_event: Callable[[BuildEvent], None] | None = None,
) -> Version

Asynchronous form of monitor() for applications that already run an event loop.

Parameters

timeout float | None: Maximum wait in seconds. None waits without a deadline.
interval float, optional: Seconds to wait between status checks. Defaults to 2 seconds.
on_event callable | None, optional: Function called once for each newly received build-log event. Omit it when no live log handling is needed.

Returns

Version: The completed version snapshot. Failed builds raise CustomToolBuildFailedError, and an expired deadline raises CustomToolBuildTimeoutError.

View fields

version str: Public version label, such as v3; use it as version when submitting a job.
status VersionStatus: Build state: Queued, Running, Complete, or Stopped.
terminal bool: Whether the build has reached a final state and will not run again.
origin str: How this version was created, such as a source build or image reuse.
created_at str: ISO 8601 creation timestamp.
started_at str: ISO 8601 timestamp when build processing began.
completed_at str | None: ISO 8601 completion timestamp, or None while the build is active.
error BuildError | None: Structured failure code and message when the build stops unsuccessfully.
tool_name str: Name of the Custom Tool that owns this version.

Raises

CustomToolBuildFailedError
CustomToolBuildTimeoutError
APIError

Examples

1version = await build.version.monitor_async(timeout=1800)

version.cancel

version.cancel(*, if_unchanged: bool = False) -> Version

Requests cancellation of an active build. By default, the SDK protects the tool lifetime without treating normal build progress as a conflict. Set if_unchanged to True to require the exact version snapshot to remain unchanged.

Parameters

if_unchanged bool, optional: False sends the SDK-managed tool-lifetime validator. True sends the exact version validator and rejects a status or revision change since this object was fetched.

Returns

Version: The version snapshot returned after Tamarind accepts the cancellation request, including its updated status.

View fields

version str: Public version label, such as v3; use it as version when submitting a job.
status VersionStatus: Build state: Queued, Running, Complete, or Stopped.
terminal bool: Whether the build has reached a final state and will not run again.
origin str: How this version was created, such as a source build or image reuse.
created_at str: ISO 8601 creation timestamp.
started_at str: ISO 8601 timestamp when build processing began.
completed_at str | None: ISO 8601 completion timestamp, or None while the build is active.
error BuildError | None: Structured failure code and message when the build stops unsuccessfully.
tool_name str: Name of the Custom Tool that owns this version.

Raises

CustomToolBuildNotInProgressError
StaleCustomToolError
APIError

Examples

1cancelled = version.cancel()

version.publish

version.publish() -> CustomTool

Makes this completed version the tool's default for submissions that omit version.

Returns

CustomTool: The updated tool snapshot. default_version identifies the version now used when a submission omits version.

View fields

name str: Stable tool name used as type when submitting jobs.
display_name str: Human-readable name shown in the UI.
description str: Explanation of what the tool does.
functions tuple[str, ...]: Short capability labels used for discovery.
status CustomToolStatus: Current lifecycle state: Draft, Building, or Deployed.
gpu_type str: GPU allocated to each run, or None for CPU-only execution.
memory str: Memory allocated to each run, expressed in GiB.
cpu int: Number of CPU cores allocated to each run.
home_disk_gi int: Writable home-disk capacity in GiB.
max_runtime_seconds int | None: Maximum runtime for a job, or None when no tool-specific limit is set.
has_source bool: Whether source has been uploaded for the current draft.
published bool: Whether the tool has a published default version.
auto_publish bool: Whether successful builds become the default automatically.
est_time str: Estimated runtime in H:M:S format.
paper_url str: Related publication URL, when configured.
tags tuple[str, ...]: Search and discovery tags.
default_version str | None: Published version name used when a job omits version.
created_at str: ISO 8601 creation timestamp.
updated_at str: ISO 8601 timestamp of the latest update.
can_edit bool: Whether the current credentials may change this tool.
can_build bool: Whether the current credentials may build this tool.
version Version | None: Requested or latest version, or None when the tool has never been built.

Raises

CustomToolNotDeployableError
StaleCustomToolError
APIError

Examples

1published_tool = version.publish()
2print(published_tool.default_version)

Return types

Page[T]: One page of SDK resources and the cursor used to request the next page.
CustomTool: Tool metadata, permissions, publication state, and runtime resources.
BuildResult: The outcome of a build request and the resulting durable version.
Version: A durable, addressable build of one Custom Tool.
ValidationReport: Local source validation result with blocking errors and non-blocking warnings.
BuildLogPage: One page of build events plus the build's current state.
CustomToolTestJob: Receipt returned immediately after submitting an unpublished version for testing.

Status values

Draft: The tool exists but has no deployed version.
Building: At least one version build is active.
Deployed: The tool has a completed deployable version.
Queued: The version build is waiting to start.
Running: The version build is running.
Complete: The version can be tested and published.
Stopped: The build failed, timed out, or was cancelled; inspect Version.error and logs().

Errors

AuthError: Authentication is missing or invalid.
APIError: The API returned an unexpected HTTP error, including authorization failures.
ValidationError: Arguments, source files, or test settings are invalid.
CustomToolNotFoundError: The requested tool or version is unavailable to this account.
CustomToolExistsError: create() used a tool name that already exists.
CustomToolUploadError: The source archive could not be prepared or uploaded.
CustomToolNotDeployableError: The selected version is not ready to test or publish.
CustomToolBuildFailedError: monitor() reached a failed terminal build state.
CustomToolBuildTimeoutError: monitor() exceeded its timeout.
CustomToolBuildNotInProgressError: cancel() targeted a build that is already terminal.
CustomToolBuildInProgressError: An operation cannot proceed while a build is active.
StaleCustomToolError: The tool changed after this SDK object was fetched; fetch it again and retry intentionally.