Skip to content
Get packs

TimesFM Time Series Forecasting

Forecast any numeric time series with Google's TimesFM model, with prediction intervals and no training

Source Download skill
Repository
google-research/timesfm
Stars
35,004
Licence
Apache-2.0
Files
15 files

What it does

Teaches an AI agent to predict future values of any series ordered in time, such as sales, sensor readings, prices or temperatures, using Google's pretrained TimesFM model without training anything yourself. The agent is guided to first run a system checker that estimates whether your machine has enough memory, then to load the model and return point forecasts with quantile prediction intervals. The folder also includes examples for anomaly detection, covariate (extra input variable) forecasting, LoRA fine-tuning and a global temperature forecast.

Great for

  • Analysts who need quick forecasts of sales, demand or sensor data without building custom models
  • Developers prototyping time series forecasting with a foundation model
  • Researchers comparing a pretrained model against classical methods like ARIMA or ETS

Skip it if

  • Safety-critical or medical decisions based on forecasts, since outputs are statistical estimates and the skill's own examples are illustrative, not validated for those uses
  • Tabular or non-temporal data, or classification and clustering tasks, where the skill itself recommends other libraries
  • Machines with little RAM or no spare disk space, since the model download is about 800 MB and the checker may block loading

Try asking

  • “Forecast the next 24 hours of electricity demand from this CSV of hourly readings”
  • “Predict next quarter's weekly sales per store with prediction intervals”
  • “Flag unusual values in my sensor log using the forecast's 90% interval”
  • “Batch-forecast 500 product demand series and export the results to a CSV”
Step-by-step guide: how to use TimesFM Time Series ForecastingInstall, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 15 original files, the Apache-2.0 licence and an attribution note.

    timesfm-forecasting.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip timesfm-forecasting.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip timesfm-forecasting.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 3

    Ask for the task

    No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.

Get better results

  • Run the bundled system checker before the first model load, as the skill instructs, and pass your real series count and context length for a memory estimate
  • Use TimesFM 2.5 rather than the archived 1.0 or 2.0 checkpoints unless you have a specific reason, since the skill says it is smaller and supports longer context
  • Review the scripts before running them; the examples write output files and the skill references scripts/check_system.py, which was not among the files reviewed

What's inside

  • examples/anomaly-detection/detect_anomalies.py
  • examples/anomaly-detection/output/anomaly_detection.json
  • examples/covariates-forecasting/demo_covariates.py
  • examples/covariates-forecasting/output/covariates_metadata.json
  • examples/covariates-forecasting/output/sales_with_covariates.csv
  • examples/finetuning/finetune_lora.py
  • examples/finetuning/README.md
  • examples/global-temperature/generate_animation_data.py
  • examples/global-temperature/generate_gif.py
  • examples/global-temperature/generate_html.py
  • examples/global-temperature/output/forecast_output.csv
  • examples/global-temperature/output/forecast_output.json
  • examples/global-temperature/README.md
  • examples/global-temperature/run_example.sh
  • SKILL.md

Redistributed unmodified from google-research/timesfm under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.