Every premium and members-only resource for $9.99/month, as a no-commitment subscription. See the plans

PalateKit

Sign in
Menu

Request a resource

Describe the resource you are missing in one sentence. We will email you when it is published.

0 / 280

A page, a video or an image online.

Unlock PalateKit

Unlock everything

Plan

Paying means accepting our terms of sale.

Secure payment by Stripe. Receipts and invoices sent by email.

Welcome back

Sign in to install, download or save this resource. A free account is enough for free resources.

Processing datasets larger than memory v1.0.0

Use this resource to process or move a dataset that exceeds memory on a modest machine

New

  • 331 downloads
  • 237 likes
  • 237 saves

Free

Install with the CLI

In a terminal, from the folder you want:

npx --yes palatekit@latest install out-of-core-data-processing

With no option, the skill goes to ~/.agents/skills/; add --agent claude-code (or codex, cursor…) to target an agent.

First time: connect the CLI with

npx --yes palatekit@latest login

The agent designs large-data processing against rules on streaming, sequential access, Parquet or CSV formats, batching, spill to disk and zstd or lz4 compression. It explains what limits throughput and orders the optimisations. You get a job that no longer runs out of memory.

Mostly picked by

Updated November 2, 2026

Card for the “Processing datasets larger than memory” skill

Details

Works in
Install guides
Generates with
Your agent's model
Licence
MIT

Similar resources

View all

What you get

  • references/: cards on batching, keyset pagination and buffered writes
  • references/: cards on codecs, dictionaries and binary protocols

Good fit for

  • A pandas script loads a multi-gigabyte CSV and runs out of memory
  • You move millions of rows on a small server

Not for

    Tags

    • data
    • streaming
    • memory
    • parquet
    • batch processing

    About

    You say

    “My script loads an 8 GB CSV with pandas and crashes. Rewrite it to read in chunks and write Parquet.”

    Card for the “Processing datasets larger than memory” skill

    Skill published by pproenca under the MIT license, delivered in its original version.

    Show more

    How to use it

    1. Install the skill in your agent's skills folder
    2. Ask your agent to process or move a dataset that exceeds memory on a modest machine
    3. Review and adjust the result

    Content preview

    Files

    • AGENTS.md10.9 KB
    • assets/templates/_template.md1.7 KB
    • LICENSE1.1 KB
    • metadata.json1.5 KB
    • NOTICE-PALATEKIT.md0.5 KB
    • references/_sections.md2.6 KB
    • references/batch-coalesce-writes-with-buffered-output.md3.4 KB
    • references/batch-keyset-pagination-over-offset.md3.5 KB
    • references/batch-pick-chunk-size-by-memory-budget.md3.1 KB
    • references/batch-process-with-stable-iterators.md3.1 KB
    • references/batch-use-vectorized-apis-not-row-loops.md2.9 KB
    • references/codec-dictionary-encoding-for-repetitive-strings.md3.5 KB
    • references/codec-prefer-binary-protocols-over-json-for-rpc.md3.6 KB
    • references/codec-train-a-zstd-dictionary-for-many-small-payloads.md3.5 KB
    • references/codec-zstd-or-lz4-as-defaults-not-gzip.md3.1 KB
    • references/conc-asyncio-for-many-network-streams-not-for-cpu.md3.8 KB
    • references/conc-overlap-compute-with-prefetch.md3.6 KB
    • references/conc-thread-pools-for-blocking-io-libraries.md3.5 KB
    • references/conc-tune-parallelism-to-the-bottleneck.md3.7 KB
    • references/fmt-avoid-deeply-nested-json-for-hot-paths.md2.7 KB
    • references/fmt-columnar-for-analytical-scans.md2.4 KB
    • references/fmt-line-delimited-for-streaming-row-ingest.md2.4 KB
    • references/fmt-prefer-schema-on-write-when-possible.md2.9 KB
    • references/fmt-push-predicates-into-the-reader.md3 KB
    • references/io-async-for-many-concurrent-streams.md2.7 KB
    • references/io-batch-and-pipeline-network-roundtrips.md3 KB
    • references/io-buffer-explicitly-for-small-records.md2.4 KB
    • references/io-mmap-for-random-or-shared-large-files.md2.8 KB
    • references/io-prefer-sequential-over-random.md3 KB
    • references/io-stream-http-bodies-with-iter-content.md4.4 KB
    • references/io-zero-copy-when-moving-bytes-as-is.md3 KB
    • references/mem-bound-the-working-set.md2.9 KB
    • references/mem-prefer-generators-over-lists-for-pipelines.md2.2 KB
    • references/mem-release-references-explicitly.md2.8 KB
    • references/mem-shrink-dtypes-before-loading.md2.1 KB
    • references/mem-stream-dont-slurp.md2.2 KB
    • references/mem-use-views-not-copies.md2.4 KB
    • references/obs-instrument-throughput-rows-per-second.md3.7 KB
    • references/obs-measure-iowait-not-just-cpu.md3.3 KB
    • references/obs-profile-with-py-spy-or-strace-for-syscall-storms.md4.6 KB
    • references/pipe-apply-backpressure-from-slow-stages.md3.7 KB
    • references/pipe-checkpoint-progress-for-resumability.md4.1 KB
    • references/pipe-prefer-pull-iteration-over-push-callbacks.md3.4 KB
    • references/pipe-use-bounded-queues-for-producer-consumer.md3 KB
    • references/spill-external-merge-sort-when-data-exceeds-ram.md3.7 KB
    • references/spill-partition-by-hash-for-out-of-core-groupby-join.md3.5 KB
    • references/spill-use-engines-that-spill-automatically.md3.6 KB
    • references/spill-use-temp-files-not-process-memory.md3.2 KB
    • SKILL.md13.3 KB

    Origin and licence

    Listed unmodified

    Author
    pproenca
    Source
    github.com/pproenca/dot-skills/tree/cf93c57cac89d6fc3e4194686000411567f5caf3/skills/.experimental/io-bound-data-processing
    Licence
    MIT

    Notice

    « io-bound-data-processing » (pproenca, MIT) livré dans sa version d'origine : fiche en français et en anglais par PalateKit.

    User reviews

    –

    No ratings yet

    Loading reviews…

    Delete your review?

    Your rating and comment will be removed from the page.