Controlled Mutation Layer (CML)

Enforced boundary for authorized state mutation

Python SDK

Current stable preview: v0.1.4

The Controlled Mutation Layer (CML) Python SDK provides a minimal, structured interface for emitting Turn records at the authority boundary.

When authoritative state changes, the SDK ensures that each mutation is:

• authorized
• explicit
• inspectable
• replayable

Each mutation produces a structured Turn record describing why the change was authorized and executed.

Why

Modern systems mutate authoritative state:

• refunds are issued
• accounts are frozen
• cases are escalated
• thresholds change

When behavior shifts, engineers ask:

• Was it model drift?
• Policy drift?
• Threshold adjustment?
• Signal change?

Most systems cannot answer these questions in a single query.

CML instruments the authority boundary so every consequential change becomes explicit and traceable.

Atomic Unit: Turn

A Turn is the atomic record of an authorized mutation.

It captures:

• turn_id - unique mutation identifier
• timestamp - UTC ISO8601 string
• pre_state - state before mutation
• signals - bounded context describing the decision
• policy_version - governing policy version
• decision - structured decision label
• post_state - state after mutation

If you cannot answer "Why did this change?", you do not have a valid Turn.

Install

From GitHub (recommended)

pip install --no-cache-dir "cml @ git+https://github.com/controlled-mutation-layer/sdk-python.git@v0.1.4"

Verify installation:

python -c "import importlib.metadata as m; print(m.version('cml'))"

For Contributors (Editable Install)

Repository: controlled-mutation-layer/sdk-python

git clone https://github.com/controlled-mutation-layer/sdk-python.git
cd sdk-python
python -m venv .venv
source .venv/bin/activate
pip install -e .
pytest -q

First Run (2 minutes)

python3 -m venv .venv
source .venv/bin/activate
pip install --no-cache-dir "cml @ git+https://github.com/controlled-mutation-layer/sdk-python.git@v0.1.4"

importdatetimefromuuidimportuuid4fromcmlimportTurnturn=Turn(
turn_id=str(uuid4()),
timestamp=datetime.datetime.now(datetime.timezone.utc).isoformat(),
pre_state={"refund_policy": "v1", "max_refund": 50},
signals={
"entity": "refund_policy",
"actor": "dev:first_run",
"reason_codes": ["policy_update"],
},
policy_version="v0",
decision="update_refund_policy",
post_state={"refund_policy": "v2", "max_refund": 75},
)
print(turn.to_json())

Quickstart

• pre_state and post_state describe what changed
• signals describes why and how the change occurred

Wrap any authoritative mutation so that every state change emits a structured Turn record.

importdatetimefromuuidimportuuid4fromcmlimportTurnpre_state= {"refund_policy": "v1", "max_refund": 50}
post_state= {"refund_policy": "v2", "max_refund": 75}
turn=Turn(
turn_id=str(uuid4()),
timestamp=datetime.datetime.now(datetime.timezone.utc).isoformat(),
pre_state=pre_state,
signals={
"entity": "refund_policy",
"actor": "ops:test",
"reason_codes": ["policy_update"],
},
policy_version="v0",
decision="update_refund_policy",
post_state=post_state,
)
print(turn.to_dict())
print(turn.to_json())

Examples

Example script: examples/refund_drift_demo.py

Run:

python examples/refund_drift_demo.py

Specification

For the conceptual model and formal definition of the mutation boundary, see controlled-mutation-layer/cml-spec.