A plant is a programmable system

Define agents, register custom tools, subscribe to run events and write your own connectors. Everything the console does is an API call you can make.

Read the docs Request API access

Python SDK
REST + gRPC
Webhooks and event stream

Start a run in eleven lines

Authenticate, pick a line, submit a lot as a goal, and stream the result.

quickstart.pypython

from footeon import Client

client = Client(api_key=os.environ["FOOTEON_API_KEY"], plant="HCMC-03")

run = client.runs.create(
    line="line-b",
    model="velo-trainer-2",
    lot_size=1200,
    size_curve="EU36-46",
    materials={"hide_batch": "LH-2291", "eva_lot": "EV-1180", "adhesive": "AD-22"},
    objective="yield,bond_strength",
)

for event in run.stream():
    print(event.step, event.status, event.duration_s)

What you get back

twin.simulate_run · succeeded · 160
cut.nest_optimize · succeeded · 252
stitch.tension_control · succeeded · 668
mold.density_control · succeeded · 591
last.fit_control · succeeded · 393
bond.window_control · needs_approval · 184

Events are also delivered by webhook and to your historian.

Teach an agent your machine

If you run equipment we do not ship a connector for, register it as a tool with a schema and the agent can call it.

custom_tool.pypython

from footeon import tool

@tool(
    name="press.set_recipe",
    description="Set temperature, pressure and dwell on a compression press.",
    validated=True,        # requires human approval by default
    schema={
        "press": "str",
        "temp_c": "float[140,190]",   # range enforced at the edge
        "pressure_bar": "float[2.0,5.0]",
        "dwell_s": "int[120,320]",
    },
)
def set_recipe(press, temp_c, pressure_bar, dwell_s):
    return opcua.write(f"ns=3;s={press}.recipe",
                       {"T": temp_c, "P": pressure_bar, "D": dwell_s})

Tool contract

Ranges enforced at the edge, outside the model
Validated tools inherit approval routing
Every call is logged with arguments and result
Failures fall back to last known-good recipe

Drive a line from your terminal

The CLI wraps the same API and is what most process engineers actually use.

footeon-cliterminal

$ footeon runs create --line line-b --model velo-trainer-2 --lot 1200
run RUN-4417 created · 8 steps planned · twin simulation queued

$ footeon runs watch RUN-4417
01 twin.simulate_run      ✓ succeeded    2m40s
02 cut.nest_optimize      ✓ succeeded    4m12s   yield 87.4%
03 stitch.tension_control ✓ succeeded   11m08s   3 skips corrected
04 mold.density_control   ✓ succeeded    9m51s   0.24 g/cm³
05 last.fit_control       ✓ succeeded    6m33s   dev 0.6 mm
06 bond.window_control    ! approval     3m04s   activation 64→66 °C

$ footeon approvals grant A-88213 --note "humidity drift confirmed"
approved · agent resumed · audit entry written

Install

pip install footeon

Ships with shell completion, a local run simulator against the twin, and offline mode for air-gapped plants.

Full CLI reference →

Everything you can build against

Six integration surfaces, all versioned and all documented.

Python SDK

Agents, runs, tools, approvals, evaluations and fleet management.

REST API

Resource-oriented, cursor-paginated, idempotency keys on every write.

gRPC streaming

Low-latency run and telemetry streams for edge and historian consumers.

Webhooks

Signed events for run state, approvals, defects and model changes.

Connector SDK

Write adapters for machines we do not ship, in Python, Go or Rust.

Edge runtime API

Local API on the factory-edge unit for offline and air-gapped operation.

Core API resources

The nouns you will work with.

Primary API resources and their purpose
Resource	Purpose	Key operations
runs	A lot moving through the line	create, get, stream, cancel
agents	A bound agent on a line	create, update autonomy, deploy
tools	Machine capabilities agents may call	register, list, validate
approvals	Pending validated-parameter changes	list, grant, deny
pairs	Per-pair traceability records	get, search, export
models	Model versions per line	list, promote, rollback

The event stream is the product

Everything the console shows is derived from this stream, and so can your systems be.

webhook payloads · RUN-4417event stream

09:14:02 hide.scan(batch="LH-2291", hides=64, sensor="line-b/cam-01")
09:14:19 64 hides mapped · 1,842 defects marked · mean usable area 91.2%
09:14:20 nest.optimize(pattern="velo-trainer-2", size_curve="EU36-46", objective="yield")
09:14:33 nest v7 · leather yield 87.4% (+8.3 pts vs baseline) · solve 12.3 s
09:18:44 cutter.execute(table="CT-2", nest="v7", blade_profile="leather-1.8mm")
09:22:56 2,412 parts cut · 0 out-of-tolerance · scrap 12.6% (plant baseline 20.9%)
09:41:10 stitch.monitor(stations=14, thread_lot="TL-77")
09:52:18 3 skipped stitches detected on lateral overlay · 3 uppers re-run · 0 escapes

Every line is real, copyable text and is announced politely to screen readers.

Delivery

Signed with HMAC-SHA256
At-least-once with idempotency keys
Replayable for 30 days
Mirrored to your historian

How to build reliably on Footeon

Do

Treat runs as idempotent and replayable
Subscribe to approvals rather than polling
Pin the API version in production integrations
Test against the local twin simulator before a live line

Avoid

Writing machine setpoints outside a registered tool
Bypassing approval routing for validated parameters
Assuming cloud connectivity at the edge
Storing per-pair records outside the audit trail

API characteristics

What to design against.

99.9% control-plane API availability target
80ms p95 edge tool-call round trip
30d event replay window
24 tool families available today

Developer questions

No. Footeon is a software and factory-edge autonomy layer that sits on top of the machines you already run. We connect over OPC UA, MQTT, Modbus TCP and vendor SDKs, read sensors and vision, and write setpoints back through the controls you already trust — with approval gates on any validated parameter.

Get API access

Sandbox credentials include the twin simulator, so you can build a full integration before touching a machine.

Request API access Read the docs