abacusai.rsi_task_run_history

Classes

RsiTaskRunHistory

Run history of a recursively self-improving (RSI) daemon task

Module Contents

class abacusai.rsi_task_run_history.RsiTaskRunHistory(client, objective=None, metricName=None, direction=None, baselineValue=None, minSample=None, noiseFloor=None, lifecycle=None, totalCount=None, gradedCount=None, limit=None, offset=None, iterations=None, trend=None, sourceDeploymentConversationId=None, workdir=None)

Bases: abacusai.return_class.AbstractApiClass

Run history of a recursively self-improving (RSI) daemon task

Parameters:

  • client (ApiClient) – An authenticated API Client instance

  • objective (str) – The user’s natural-language objective the task improves at.

  • metricName (str) – The name of the measured metric.

  • direction (str) – Whether the metric should ‘increase’ or ‘decrease’.

  • baselineValue (float) – The metric value measured at task setup.

  • minSample (int) – Minimum sample size required to grade a run.

  • noiseFloor (float) – Absolute metric band treated as no change.

  • lifecycle (str) – The lifecycle of the daemon task.

  • totalCount (int) – Total number of runs with an iteration record.

  • gradedCount (int) – Number of runs whose iteration record has been graded (closed).

  • limit (int) – The page size used for the iterations list.

  • offset (int) – The offset used for the iterations list.

  • iterations (list) – One page of per-run records, newest first (iteration_index, status, value_before, value_after, delta, n_sample, verdict, change_summary, measured_at, failure_reason, run_record, created_at, run_lifecycle).

  • trend (list) – Compact full history for charting, oldest first (index, value, verdict; the in-flight run appears last with pending=true at its comparison baseline).

  • sourceDeploymentConversationId (str) – The conversation whose filesystem holds the RSI workdir (for fetching run records).

  • workdir (str) – The RSI working directory on that filesystem.

objective = None
metric_name = None
direction = None
baseline_value = None
min_sample = None
noise_floor = None
lifecycle = None
total_count = None
graded_count = None
limit = None
offset = None
iterations = None
trend = None
source_deployment_conversation_id = None
workdir = None
deprecated_keys
__repr__()
to_dict()

Get a dict representation of the parameters in this class

Returns:

The dict value representation of the class parameters

Return type:

dict