sktime-cli

The command line for sktime : built for AI agents and humans.

Discover estimators, fetch datasets, inspect time series files, and run fit / predict / evaluate workflows straight from your shell.

Python License Status Ruff Built on sktime

sktime-cli demo: registry search, datasets load, run fit, run predict

Highlights

  • Stateless, one-shot commands - every invocation is a single process: read files, call sktime, write results, exit with a meaningful code. State lives on disk only, Hugging Face CLI style.
  • Registry-native discovery - registry search filters sktime's full estimator registry by scitype and capability tags, served from a disk cache so warm searches are instant.
  • Estimator spec strings -models are named the way you'd write them in Python: "NaiveForecaster(sp=12)", with pipelines via *, ensembles via +, and multiplexers via |.
  • Any format in, any format out - csv, parquet, json, .ts, .tsf, .arff; every command speaks --format human|agent|json|quiet.
  • Agent-first contract - one JSON document on stdout, structured errors on stderr, stable exit codes, and a ready-to-drop-in agent skill.

Installation

uv tool install sktime-cli   # or: pip install sktime-cli

Verify your setup and see which optional dependencies are available:

sktime-cli doctor
sktime-cli doctor output

Quickstart

# what can I use?
sktime-cli registry search forecaster -t capability:missing_values=true
sktime-cli registry describe NaiveForecaster

# get data
sktime-cli datasets load airline --output airline.csv
sktime-cli data inspect airline.csv

# fit, predict, evaluate — estimators are given as sktime spec strings
sktime-cli run fit "NaiveForecaster(sp=12)" --data airline.csv --model-out model.zip
sktime-cli run predict --model model.zip --fh 1:12
sktime-cli run evaluate "NaiveForecaster(sp=12)" --data airline.csv --fh 1:12 \
  --metric MeanAbsolutePercentageError

Command overview

Group Commands What it does
registry search · describe · tags · types Discover sktime estimators by scitype, name, and capability tags
datasets list · describe · load Browse and fetch built-in, UCR/UEA, Monash, and fpp3 datasets
data inspect · convert · split Detect mtypes/scitypes, convert formats, temporal train/test split
run fit · predict · fit-predict · evaluate One-shot workflows for forecasting and classification
model inspect Look inside a saved model artifact; round-trip its spec
(top level) version · env · doctor · cache Environment info, health check, workspace management

See the CLI reference for every option.

Built for AI agents

sktime-cli treats agents as first-class users. Add --json to any command and you get exactly one parseable JSON document on stdout; errors are JSON on stderr with stable codes and actionable hints.

sktime-cli JSON output and structured error with exit code
exit meaning
0 success
1 library or unexpected failure
2 usage error
3 missing optional dependency (the hint says what to install)
4 estimator / dataset / model not found
5 data validation or spec error

The full agent-facing contract and task-oriented workflows live in skills/sktime-cli/SKILL.md — drop the skills/ folder into your agent's skill directory (e.g. .claude/skills/) and your agent knows how to drive the CLI. The same file ships inside the wheel.

Documentation

Document Contents
Architecture Repository layout, module map, dependency layering, data flow
Design Design decisions: state model, output contract, error model, spec engine, caching
CLI reference Full command tree with options
Agent skill The contract agents are given
Plan v0.0.1 milestones and roadmap

Relation to sktime-mcp

sktime-cli is the CLI sibling of sktime-mcp. The command vocabulary mirrors its tool names (registry search ~ query_registry, registry describe ~ describe_component, run fit/predict/evaluate), so agent knowledge transfers between the two.

Status

v0.0.1 — early alpha. Discovery and one-shot runs are complete; see PLAN.md for what's deferred to v0.0.2+. An adversarial agent benchmark suite (foundation + hard tiers, provider-neutral run records, scoring keys) is being developed on the feat/adversarial-benchmark branch.

License

BSD 3-Clause, consistent with sktime.

Project links

Data verified by PyPI on Aug 25, 2026
Data provided by the project maintainers, verified at the time the release was uploaded to PyPI.

Key dates

PyPI data
Data sourced directly from PyPI's database.
  • Released:
Newer release available (0.0.2)

1 maintainer

PyPI data
Data sourced directly from PyPI's database.
Avatar for siddharth899 from gravatar.com siddharth899

Credits

Author: Siddharth

GitHub Statistics

Data verified by PyPI on Aug 25, 2026
The GitHub source repository was provided by the project maintainers and verified by PyPI at the time of upload. Stars, forks, and open issues/PRs are derived from that repository and have not been independently verified.

License

BSD License (BSD 3-Clause License)

Requires

Python <3.15, >=3.10

Provides Extra

parquet

Tags

cli forecasting machine-learning sktime time-series

Classifiers

Report project as malware

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

sktime_cli-0.0.1.tar.gz (119.4 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

sktime_cli-0.0.1-py3-none-any.whl (40.3 kB view details)

Uploaded Python 3

File details

Details for the file sktime_cli-0.0.1.tar.gz.

File metadata

  • Download URL: sktime_cli-0.0.1.tar.gz
  • Upload date:
  • Size: 119.4 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for sktime_cli-0.0.1.tar.gz
Algorithm Hash digest
SHA256 626ddc7b5bd6c5917889456aaa108e90174ea7aa58718946bc23346fb8ad7101
MD5 62d50c74ec4b721fa5c140f9b6b4bb1e
BLAKE2b-256 f2735e5de05a6c25d565f964de182e88109c4a33648354bf5a424a296a14c5d9

See more details on using hashes here.

Provenance

The following attestation bundles were made for sktime_cli-0.0.1.tar.gz:

Publisher: release.yml on siddharth7113/sktime-cli

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file sktime_cli-0.0.1-py3-none-any.whl.

File metadata

  • Download URL: sktime_cli-0.0.1-py3-none-any.whl
  • Upload date:
  • Size: 40.3 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/7.0.0 CPython/3.13.14

File hashes

Hashes for sktime_cli-0.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 fc5afec071208be508825feb0b84ee2da26aef2578b3462c7e0af7bb3b51911f
MD5 91cce6164cf3ee4281ec1dc733d9a3c6
BLAKE2b-256 761da1f5a53a9b2615d4d5f79b707548c74f2a0003959c6450d06d0540de0a90

See more details on using hashes here.

Provenance

The following attestation bundles were made for sktime_cli-0.0.1-py3-none-any.whl:

Publisher: release.yml on siddharth7113/sktime-cli

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Release history Release notifications | RSS feed

0.0.2

2 files

This release

0.0.1 This release

2 files

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page