FluidGrids Python SDK
Official Python SDK for FluidGrids, the workflow automation platform.
It is a thin shell over burdenoff-sdk-libs,
like vibecontrols-sdk: FluidGrids operations live in the shared fluidgrids
module (sdk.fluidgrids), and the platform modules are mounted alongside
(sdk.auth, sdk.workspaces, sdk.tags, sdk.files, sdk.billing,
sdk.notifications, sdk.support, sdk.store, sdk.integrations, …).
Installation
pip install fluidgrids-sdk
pip install "fluidgrids-sdk[subscriptions]" # also websockets, to follow runs live
fluidgrids-sdk is published on PyPI and requires Python 3.10+. To work on the
SDK itself, install it from a checkout with pip install -e ".[dev,test]" (see
Development below). The full guide is at
docs.fluidgrids.ai/developers/python-sdk.
Quick start
import asyncio
from fluidgrids_sdk import FluidGridsSDK
async def main() -> None:
async with FluidGridsSDK(access_token=token, refresh_token=refresh) as sdk:
# Workspace calls need a workspace token as well as the platform token.
await sdk.auth.issue_workspace_token(workspace_id, organization_id)
page = await sdk.fluidgrids.workflows.list({"status": "ACTIVE", "limit": 20})
run = await sdk.fluidgrids.workflows.trigger(
{"workflow_key": "invoice-sync", "version": "1.0.0", "trigger_node_id": "start",
"trigger_data": {"amount": 42}}
)
done = await sdk.fluidgrids.runs.wait_for_completion(run.run_id, timeout=600)
logs = await sdk.fluidgrids.runs.logs(run.run_id, min_level="WARN")
asyncio.run(main())
Get tokens with any sdk.auth flow: device code, PKCE, or client credentials
for automation.
The FluidGrids module
| Namespace | Operations |
|---|---|
fluidgrids.workflows |
list, iterate, get(id) / get(workflow_key=, version=), versions, summary, diff_versions, create, update, delete, trigger, publish, activate, deactivate, archive, duplicate, import_template, restore_version, execute_step, export, import_, create_version, validate, interface, start, start_with_update |
fluidgrids.runs |
list, iterate, get, logs, iterate_logs, hierarchy, node_states, node_state, cancel, pause, resume, retry_node(run_id, node_id), replay, retry, run_from_node, can_run_from_node, wait_for_completion, watch, watch_node_states, stream_logs, describe, by_workflow_id, pending, count, history, history_all, signal, stack_trace, determinism, terminate, result, query, update, update_result, reset, batch.start|get|list|cancel |
fluidgrids.activities |
start, complete, fail, heartbeat (task token or run and activity id), pause, unpause, reset, update_options, pause_rules.list|create|delete |
fluidgrids.nodes |
list, iterate, get(node_type, node_version), list_custom, iterate_custom, manifest, usage_status, list_operation_configs, iterate_operation_configs, get_operation_config, list_installations, install, uninstall, set_installation_enabled |
fluidgrids.credentials |
list, iterate, get, list_types, iterate_types, create, authenticate_with_api_key, authenticate_with_basic_auth, update, delete, test |
fluidgrids.webhooks |
list, get, create, update, delete, toggle, activate, deactivate, list_deliveries, list_all_deliveries, iterate_all_deliveries, retry_delivery, list_filters |
fluidgrids.scheduled_jobs |
list, iterate, get, create, update, delete, set_enabled, pause, resume, restore, scheduler_status |
fluidgrids.dashboard |
summary, usage({"start_date", "end_date"}), error_analytics |
fluidgrids.search |
search, by_entity_type, search_runs |
fluidgrids.admin |
system_info |
fluidgrids.variables |
list, get, create, update, delete, reveal |
fluidgrids.data_tables |
list, iterate, get, create, update, delete, query_rows, iterate_rows, insert_row, update_row, delete_rows |
fluidgrids.static_data |
get, list, reset |
fluidgrids.pinned_data |
pin, unpin |
fluidgrids.tool_approvals |
list_pending, iterate_pending, get, decide |
fluidgrids.log_streams |
list, get, create, update, delete, test |
fluidgrids.secrets_providers |
list, save, test_connection |
fluidgrids.tags |
search, list_available, create, list_for_entity, replace_for_entity, tag_entity, untag_entity |
fluidgrids.usage |
time_series, detailed_node_usage, workflow_usage_summary |
Responses are snake_case dataclasses. Inputs are snake_case dicts; only their
top-level keys are converted to GraphQL names, so JSON values such as
definition_json or trigger_data are sent exactly as given.
Following a run
runs.watch, runs.watch_node_states and runs.stream_logs are async iterators over
the run's subscriptions. Each one reads the run over HTTP first, so a run that has
already finished ends the loop without opening a socket, and the events are an
accelerator on top of a poll (every 5 s by default), so a dropped socket costs time
and not updates. A run that ended as FAILED is a final update (the loop ends
normally; runs.result is the call that raises). Leaving the loop, cancelling the
task or timeout close the socket.
async for update in sdk.fluidgrids.runs.watch(run.run_id, timeout=600):
print(update.status, update.progress)
async for line in sdk.fluidgrids.runs.stream_logs(run.run_id):
print(line.level, line.message)
Subscriptions need the optional websockets package: pip install "fluidgrids-sdk[subscriptions]" (it is loaded when a stream first needs a socket;
without it a stream fails with an error that names it, and every other call is
unaffected). An https gateway is used as wss: with an http endpoint the tokens
travel in clear text, so use http only for a local test gateway.
Paging through a list
The iterate* methods walk a list page by page (page_size 50, 500 for logs and rows;
max_items and max_pages end the walk normally; a server that never stops is cut off
after 1000 pages with PAGINATION_LIMIT):
async for run in sdk.fluidgrids.runs.iterate({"workflow_key": "invoice-sync"}, page_size=100):
print(run.id, run.status)
A credential is an integration connection, created in two steps: create (no
secret material), then authenticate_with_api_key / authenticate_with_basic_auth.
Run control: updates, queries, reset, batches and activities
With the fluidgrids.run-control feature on the workspace, a durable run can be
asked questions, sent updates, reset and operated on in batches, and an activity
that waits for an asynchronous completion can be finished from outside the run:
started = await sdk.fluidgrids.workflows.start_with_update(
"order-saga", {"name": "add_item", "args": {"sku": "A1"}}, workflow_id="order-7"
)
progress = await sdk.fluidgrids.runs.query(started.run.id, "progress")
added = await sdk.fluidgrids.runs.update(started.run.id, "add_item", {"sku": "B2"})
fork = await sdk.fluidgrids.runs.reset(started.run.id, reason="hotfix deployed")
await sdk.fluidgrids.runs.batch.start(
"SIGNAL",
filter={"workflow_key": "order-saga", "status": "RUNNING"},
payload={"signal_name": "expedite"},
reason="carrier delay",
)
await sdk.fluidgrids.activities.complete(task_token, {"approved": True})
workflows.start also takes task_queue, fairness_key and fairness_weight;
a self-hosted queue needs fluidgrids.self-hosted-workers. Without the feature
the service answers each of these with FEATURE_DISABLED; calls without them
behave as before.
Standalone activities
With fluidgrids.standalone-activities on the workspace, one node operation runs
as a durable activity, with retries, timeouts, heartbeats and an event history
(the service runs it as a durable run of the read-only system workflow
system-standalone-activity):
started = await sdk.fluidgrids.activities.start(
{
"node_type": "internal:http",
"operation": "request",
"args": {"url": "https://example.com/hook", "method": "POST"},
"retry": {"maximum_attempts": 3, "initial_interval": "1s"},
"start_to_close_timeout": "30s",
"idempotency_key": "hook-2026-10-30",
}
)
output = await sdk.fluidgrids.runs.result(started.run.id)
The arguments are literal values (never template-evaluated, never encoded by the
payload codec); routing nodes such as internal:if are refused with
durable.activityTypeNotAllowed. A repeated idempotency_key from the same caller
returns the run it started (existing is true; needs fluidgrids.durable-parity).
Without the feature the service answers with FEATURE_DISABLED.
Payload codec and converter
Pass codec= (encryption, compression) and/or payload_converter= (values that
are not JSON) to FluidGridsSDK. The client then encodes the payloads it sends
(workflow input, signal payloads, update and query arguments, activity results,
failure details, heartbeat details) and decodes the ones it reads (run output,
update results, query answers, history events). Ids, names, the memo and search
attributes stay in clear text, and no platform token is given to the codec.
Configuration
| Argument / env | Default |
|---|---|
workspace_endpoint / FLUIDGRIDS_WORKSPACE_ENDPOINT |
https://graphqlworkspaces.burdenoff.com/workspaces/graphql |
global_endpoint / FLUIDGRIDS_GLOBAL_ENDPOINT |
https://graphql.burdenoff.com/global/graphql |
product_origin / FLUIDGRIDS_ORIGIN |
https://app.fluidgrids.ai (prod), https://alphaapp.fluidgrids.ai (alpha) |
The workflow service is product-scoped: the gateway works out the product from
the request Origin, which httpx does not send by itself. The SDK sets it for
the production and alpha gateways; pass product_origin for a custom or local
gateway.
Development
python -m venv .venv && .venv/bin/pip install -e ".[dev,test]"
make sanity
tests/test_graphql_documents.py validates every FluidGrids document against the
vendored schema/wspace.graphql (see schema/README.md to refresh it). Change
FluidGrids behaviour in sdk-libs-python (modules/fluidgrids/) and its Node twin;
this repo only picks modules and product defaults.
License
Proprietary — Copyright (c) Algoshred Technologies Private Limited
Metadata
Release files for fluidgrids-sdk 2026.1117.2
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | Provenance | Actions |
|---|---|---|---|---|
| fluidgrids_sdk-2026.1117.2.tar.gz | 1.5 MB | Without provenance | Details for fluidgrids_sdk-2026.1117.2.tar.gz |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Actions Reset |
|---|---|---|---|---|
| fluidgrids_sdk-2026.1117.2-py3-none-any.whl | Python 3 | none | any | Details for fluidgrids_sdk-2026.1117.2-py3-none-any.whl |
Total release size: 1.5 MB
Release files / fluidgrids_sdk-2026.1117.2.tar.gz
| Download URL | fluidgrids_sdk-2026.1117.2.tar.gz |
|---|---|
| Size | 1.5 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
7ef3378bf5f801e035bd10a05cfdc7b5ef5c92e7b40362a9a3ed053b42c6a6d8
|
|
BLAKE2b-256 checksum How to use checksums |
8992ddb06e2417991c388d0500f9a747d7fc4fc842eeb1b96ee4d161477e204e
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Release files / fluidgrids_sdk-2026.1117.2-py3-none-any.whl
| Download URL | fluidgrids_sdk-2026.1117.2-py3-none-any.whl |
|---|---|
| Size | 10.9 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
889d6d2c443128e9d5c32fb34ee834f7c84d7ede7e3d5012c9e236f90e99d401
|
|
BLAKE2b-256 checksum How to use checksums |
92f6d36298f6fe4e2418d78387c09db04dd1a27c4e60105c146cd33c033282eb
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
No |
| Uploaded via |
twine/7.0.0 CPython/3.13.16
|
Security for fluidgrids-sdk 2026.1117.2
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
No provenance found
None of the files in this release were uploaded with provenance metadata.
A note on package security
PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.
To learn more about protecting yourself from malicious actors, visit our package security documentation.