Skip to content
Skills Library
Data & analytics · Building & coding

time-series-data

/time-series-data

Store time-series data with retention downsampling, compression, out-of-order handling, and the right store choice. Use when handling metrics, events, or sensor streams that grow relentlessly by time.

Repository
AI-SKILLS ↗
Licence
MIT
Last repository change
05.09.2026

Assessment and security

90/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content25 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 05.09.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Store time-series data with retention downsampling, compression, out-of-order handling, and the right store choice. Use when handling metrics, events, or sensor streams that grow relentlessly by time.

Original author description

Store time-series data with retention downsampling, compression, out-of-order handling, and the right store choice. Use when handling metrics, events, or sensor streams that grow relentlessly by time.

How it works

  1. Method
  2. Boundaries
Original SKILL.md headings · 76 lines, 445 words
SKILL.md preview
---
name: time-series-data
description: Store time-series data with retention downsampling, compression, out-of-order handling, and the right store choice. Use when handling metrics, events, or sensor streams that grow relentlessly by time.
---

# Time-series data

Time-series data (metrics, events, sensor readings) has a defining
shape: append-heavy, time-ordered, immense volume, and value that
decays with age. A general-purpose table handles it badly; the
techniques are all about exploiting the time dimension for storage,
retention, and query.

## Method

1. **Partition by time.** Range-partition on the timestamp
   (daily or hourly by volume: see data-partitioning): so
   queries filtering by time range prune to a few
   partitions, and old data drops by detaching a partition
   (near-instant) instead of deleting billions of rows.
   Time partitioning is the foundational move that makes
   everything else efficient.
2. **Downsample and expire by age.** Recent data at full
   resolution (per-second), older data aggregated
   (per-minute, then per-hour), oldest expired entirely:
   because nobody queries last year at second resolution.
   Automated retention policies (roll up then drop raw)
   keep storage bounded and queries fast (see
   data-retention's tiering, sensor-data-handling for the
   edge side). This aggregate-then-drop is the standard
   time-series lifecycle.
3. **Exploit compression.** Time-series compresses
   extremely well (timestamps are near-regular, values
   change slowly: delta and delta-of-delta encoding, plus
   columnar storage): purpose-built stores achieve 10x+
   ratios. Use the store's native compression; storing
   time-series uncompressed in a general table wastes an
   order of magnitude of disk.
4. **Handle out-of-order and late data.** Readings arrive
   late (network delays, batched device uploads: see
   iot-messaging, sensor-data-handling); the store and
   queries must accept writes into past windows and
   recompute affected rollups (see incremental-processing's
   late-data window). A design assuming strictly-increasing
   timestamps breaks on the first delayed batch.
5. **Choose the right store for the scale.** Purpose-built
   time-series databases (Timescale, InfluxDB,
   Prometheus-class) give partitioning, compression,
   downsampling, and time-oriented query functions out of
   the box; a relational table with manual partitioning
   works at modest scale. Match the store to the volume and
   query needs (see managed-vs-selfhosted): at high ingest
   rates, the specialized store's built-in mechanics beat
   hand-rolling them.
6. **Model for time-range queries and aggregation.** The
   dominant queries are "values over a time range,
   aggregated by interval and grouped by series/tag":
   design tags/dimensions for the group-bys you need (like
   warehouse-modeling's grain, for time), index the time
   plus tag columns (see indexing-strategy), and precompute
   common rollups (continuous aggregates: see
   materialized-views). Point lookups are rare;
   range-scan-and-aggregate is the workload.

## Boundaries

- Time-series stores optimize for append-heavy time-ordered
  data; they are poor for data needing updates, complex
  relations, or transactions (see transactions-isolation):
  do not force relational workloads into them or vice
  versa.
- Metrics/observability time-series (see
  infrastructure-monitoring) and business/analytics
  time-series overlap in technique but differ in tooling
  and retention needs; match the stack to the use.
- Extremely high-cardinality tag combinations blow up many
  time-series stores (each unique series is tracked);
  cardinality management is the specific scaling gotcha to
  design against.

Versions

The database has no other skills named time-series-data.

Database snapshot: 13.11.2026 · package version 4bfea1161371 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100influxdb
    Expert InfluxDB time-series assistance covering Line Protocol, Flux/InfluxQL, retention policies, downsampling, and bucket management. Use when storing IoT metrics, application telemetry, or financial time-series data.
  • 90time-series-analysis
    Analyze and forecast time-ordered data respecting trend, seasonality, autocorrelation, and the arrow of time. Use when working with data indexed by time: metrics, sales, sensor readings, or any forecast.
  • 100causal-ts-discovery
    Discover causal structure in time-series data with the causal-ts library. Use when the user wants a causal graph, or lead-lag / temporal dependencies, from time-series, sensor, or sequential data. NOT for effect size, ATE, or counterfactuals — that is effect estimation (see the handoff at the end).
  • 100forecasting-multiple-series
    Forecasts multiple time series simultaneously using a global model with ForecasterRecursiveMultiSeries or ForecasterDirectMultiVariate. Covers data formats, encoding, per-series transformers, and multi-series backtesting. Use when the user has two or more related time series.
  • 100setup-timescaledb-hypertables
    Use this skill when creating database schemas or tables for Timescale, TimescaleDB, TigerData, or Tiger Cloud, especially for time-series, IoT, metrics, events, or log data. Use this to improve the performance of any insert-heavy table. **Trigger when user asks to:** - Create or…
  • 100forecasting-time-series-data
    Process this skill enables AI assistant to forecast future values based on historical time series data. it analyzes time-dependent data to identify trends, seasonality, and other patterns. use this skill when the user asks to predict future values of a time ser... Use when…

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.