跳转到正文

robot_control API ​

完整接口目录 · 插件运行时用法 · 工具包下载

模块说明 ​

text
Prepare native robot observations and bounded control operations without IO.

RobotControl ​

python
class RobotControl

SDK 说明:

text
Explicit intent builders over the existing managed Host capabilities.

Persist each intent before submitting through Operations. Host validation,
grants, attachment checks and the shared world gate remain authoritative.
These builders never submit, retry, refresh observations or interpret a
receipt. Preserve unknown outcomes and reconcile their original operation.

RobotControl.__init__ ​

python
def __init__(self, operations: Operations)

RobotControl.inventory_support ​

python
def inventory_support(self, *, idempotency_key: str, deadline: str) -> OperationIntent

SDK 说明:

text
Read the implementation support matrix, not inventory evidence.

RobotControl.equipment_move ​

python
def equipment_move(
    self,
    source_window: int,
    source_slot: int,
    destination_window: int,
    destination_slot: int,
    count: int,
    *,
    expected_source_stack_id: int,
    expected_destination_stack_id: int,
    idempotency_key: str,
    deadline: str,
) -> OperationIntent

SDK 说明:

text
Move an original stack between main and observed-empty equipment.

Windows: main 0 (slots 0..35), offhand 119 (slot 0), armour 120
(slots 0..3). Exactly one endpoint must be main; armour count is 1.
Destination stack ID must be 0. Server ACK is not independent readback.

RobotControl.action_audit ​

python
def action_audit(
    self,
    *,
    idempotency_key: str,
    deadline: str,
    after: Mapping[str, Any] | None = None,
) -> OperationIntent

SDK 说明:

text
Read local action audit; after retains connection_id and sequence.

Audit records describe local dispatch, not server-confirmed world effects.

RobotControl.death ​

python
def death(
    self,
    *,
    idempotency_key: str,
    deadline: str,
    after: Mapping[str, Any] | None = None,
    limit: int = 8,
) -> OperationIntent

SDK 说明:

text
Read retained self death evidence with a connection_id/revision cursor.

RobotControl.auth_input_observe ​

python
def auth_input_observe(
    self,
    *,
    idempotency_key: str,
    deadline: str,
    connection_id: str | None = None,
    max_age_ms: int = 1000,
) -> OperationIntent

SDK 说明:

text
Read retained typed input evidence; max_age_ms is 1..10000.

A read does not refresh packet age or establish the server's tick clock.

RobotControl.action_guard ​

python
def action_guard(
    self,
    *,
    idempotency_key: str,
    deadline: str,
    connection_id: str = '',
    max_age_ms: int = 500,
    retained_fence: dict | None = None,
) -> OperationIntent

SDK 说明:

text
Prepare existing sequential death/inventory/input reads, not atomic state.

RobotControl.auth_input_bootstrap ​

python
def auth_input_bootstrap(
    self,
    *,
    connection_id: str,
    expected_auth_input_revision: int,
    expected_uuid: str,
    dimension: int,
    position: Sequence[float],
    pitch: float,
    yaw: float,
    head_yaw: float,
    candidate_tick: int,
    auth_input_height_offset: float,
    idempotency_key: str,
    deadline: str,
) -> OperationIntent

SDK 说明:

text
Prepare one explicit neutral candidate using the existing bootstrap API.

RobotControl.jump ​

python
def jump(
    self,
    *,
    connection_id: str,
    initial_tick: int,
    position: Sequence[float],
    ground_y: float,
    eye_height: float,
    pitch: float,
    yaw: float,
    head_yaw: float,
    delta_semantics: str,
    dimension: int,
    idempotency_key: str,
    deadline: str,
    expected_auth_input_revision: int | None = None,
) -> OperationIntent

SDK 说明:

text
Prepare the Host's bounded 24-frame jump with packet 144 permission.

Caller supplies grounded evidence and exact input coordinates; position Y
must equal ground_y + eye_height. delta_semantics is end_velocity or
position_difference. Successful dispatch does not prove landing.

RobotControl.move ​

python
def move(
    self,
    target: Sequence[float],
    *,
    connection_id: str,
    initial_tick: int,
    dimension: int,
    auth_input_height_offset: float,
    idempotency_key: str,
    deadline: str,
    speed: float = 0.1,
    tolerance: float = 0.2,
    timeout_ms: int = 5000,
    expected_auth_input_revision: int | None = None,
    experimental_ground: Mapping[str, Any] | None = None,
) -> OperationIntent

SDK 说明:

text
Prepare the native Host movement loop; do not start it or create a key.

Uses NativeMovement's wire contract and bounded uint64 checks. Ground
mode requires an exact revision and explicit initial evidence/budgets.
Packet 144 permission and Host world gate still apply. Movement evidence
is retained as reported; modeled_target_observed is only dispatched.

RobotControl.auth_input_sequence ​

python
def auth_input_sequence(
    self,
    *,
    connection_id: str,
    expected_uuid: str,
    initial_tick: int,
    expected_auth_input_revision: int,
    initial_position: Sequence[float],
    bounds_min: Sequence[float],
    bounds_max: Sequence[float],
    dimension: int,
    cadence_ms: int,
    timeout_ms: int,
    auth_input_height_offset: float,
    frames: Sequence[Mapping[str, Any]],
    idempotency_key: str,
    deadline: str,
) -> OperationIntent

SDK 说明:

text
Prepare explicit wire frames without synthesizing tick or input state.

Host accepts 4..12 frames, with the last four neutral, fixed rotation and
Y, consecutive ticks and only InputFlagUp. Each frame retains all exact
PascalCase wire keys. Cadence is 50..100 ms, timeout 100..1500 ms, and
each axis bound spans at most 8 blocks. Packet 144 grant is required.
Success is local admission, not verified walking or server-side stopping.

RobotControl.operations ​

python
self.operations

公开实例属性,在 __init__ 中设置。

Neomega · 使用、开发与参考文档