Marivo 0.4.10

Marivo 0.4.10 strengthens multi-metric and temporal analysis, makes quality evidence more precise, and improves the handoff from semantic authoring to analysis for agents.

Highlights

Multi-metric observations now preserve each metric binding’s aggregation semantics across transforms and only expose reaggregation when the metric is additive. Mixed temporal axes provide a ready-to-copy split plan instead of leaving the agent to infer how to separate the observations.
Temporal readiness now surfaces an undeclared timezone for inferred naive time axes, so the author can make the time interpretation explicit before execution. Cumulative observations over decimal inputs are normalized to a stable float64 representation in dense execution paths.
Semantic authoring can hand off to analysis automatically once the semantic object is ready. Session recovery now identifies the exact session-owned state, while the analysis bootstrap and environment guidance make the supported entry path and its traceability clearer.
Quality assessment distinguishes correctness-blocking issues from contextual evidence, and multi-metric evidence digests retain the input metric order. Derived metrics also expose their details through the semantic catalog.

Changed

Quality checks and readiness reports now carry more precise temporal, metric-binding, and evidence scope information into repair guidance.
Analysis continuation and semantic handoff guidance use the current public workflow and explicit session identity instead of relying on an implicit latest result or a compatibility-only quality field.

Fixed

Multi-metric evidence no longer changes the caller’s metric order while building a digest or summary.
Per-binding aggregation metadata is preserved when metric values are loaded, transformed, and reused by analysis operations.
Decimal cumulative values no longer remain in an inconsistent numeric type on dense paths.
Inferred naive time axes now report the missing timezone authority rather than silently proceeding with an ambiguous interpretation.
Mixed-axis multi-metric observations now return a concrete split plan that can be copied into the next operation.

Documentation

The analysis and semantic guides document the automatic handoff, explicit session recovery identity, correctness-only quality blocking, temporal authority, multi-metric value semantics, and derived-metric catalog details.

Installation

curl -fsSL https://marivo.io/install.sh | bash

In mainland China, use:

curl -fsSL https://marivo.io/install-marivo-cn.sh | bash

For manual installation:

pip install marivo

Install the backend extra for each datasource you plan to use. Marivo 0.4.10 requires Python 3.10 or newer.