skforecast-ai

Official

Overview Schema Source Related Servers Discussions Admin

create_cv

Build a time-series cross-validation strategy for a forecasting plan, configure folds, gaps, refit, and training size, then get a CV id for backtesting.

Instructions

Build a time series cross-validation strategy (TimeSeriesFold) for a plan. The response states its cost in cost. Returns a cv id.

Input Schema

TableJSON Schema

Name	Required	Description
`gap`	No	Observations between the end of training and the test set. Null for 0. A direct forecaster (ForecasterDirect, ForecasterDirectMultiVariate) cannot be backtested with a gap.
`refit`	No	Whether to train again in every fold (true), never (false) or every n folds (an integer). Null for false: train once. Refitting multiplies the cost; ForecasterStats is refitted in every fold whatever it says.
`plan_id`	Yes	Id of the plan to backtest (from `plan` or `refine_plan`).
`skip_folds`	No	Folds are numbered from 0, and fold 0 always runs. An integer n keeps folds 0, n, 2n, ...; a list skips the folds at those numbers (each at least 1 and below the number of folds). Null for every fold.
`fold_stride`	No	Observations the test set advances between folds. Null for `steps` (back-to-back folds); a larger value means fewer folds.
`fixed_train_size`	No	Whether the training window keeps its size when refitting (true) or grows (false). Null for false. Only with `refit` true or an integer: with `refit` false or null it is an error, as it would not run (ForecasterStats aside).
`initial_train_size`	No	Observations of the first training set (a number), or the ISO 8601 date that ends it ('2005-06-01'). Null for the default: 70 % of the series, at least what the lags need and leaving room for two folds, written as a date when the data has dates.
`allow_incomplete_fold`	No	Whether the last fold may have fewer than `steps` observations. Null for true.

Output Schema

TableJSON Schema

Name	Required	Description	Default
`id`	Yes
`cost`	No
`kind`	Yes
`files`	No
`links`	No
`notices`	No
`summary`	Yes
`changeable`	No
`notices_omitted`	No
`values_included`	No
`summary_truncated`	No

Schema Changelog

Changes observed during successful MCP inspections.

First observedv0.1.0Oct 15, 2026

TDQS

B3.2/5.0

Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare destructiveHint=false and readOnlyHint=false, covering the mutation/safety profile. The description adds that the response reports its own 'cost' and returns a cv id, which is real behavioral context. It says nothing about persistence, lifetime, or whether the strategy can be modified or deleted afterwards.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Three short sentences, front-loaded with the purpose, no filler. The final sentence 'Returns a cv id' is largely duplicative of the output schema, which costs it slightly.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

With an output schema present, return values need not be explained, and the schema documents all eight parameters thoroughly. The definition is complete enough to invoke correctly; only the relationship to sibling tools is unaddressed.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, and the eight parameters have detailed inline descriptions covering gap, refit, fold_stride, skip_folds, and initial_train_size semantics. The description adds nothing about parameters, so baseline 3 applies; the schema carries the full parameter burden.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

Names a specific verb and artifact: 'Build a time series cross-validation strategy (TimeSeriesFold) for a plan.' An agent can tell it creates a CV configuration rather than running a backtest. It doesn't explicitly differentiate from the sibling 'backtest' or 'forecast', so it stops short of a 5.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines2/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

No statement of when to use this tool versus alternatives, and no prerequisites. With siblings like 'backtest' present, the description never says whether this must be created first or how it relates to them. Usage must be inferred entirely.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Deploy Server