chunking-strategies SKILL.md skillA

chunking-strategies from amey-thakur/ai-skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add amey-thakur/ai-skills/chunking-strategies@git:20260728.fe84bd4

Pins git:20260728.fe84bd4 in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
amey-thakur/ai-skills
Path
skills/ai-memory-rag/chunking-strategies/SKILL.md
Versions recorded
1
Latest version
git:20260728.fe84bd4, committed Jul 28, 2026
Size
2,162 bytes
Repository stars
14
License
MIT

Split documents into retrieval units that preserve meaning, so a retrieved chunk answers the question rather than trailing off mid-thought. Use when building or fixing a retrieval pipeline whose results are technically relevant but useless.

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# Chunking strategies

Chunking decides what retrieval can ever return. A chunk that splits a
definition from its example, or a table from its header, cannot be
rescued by a better embedding model or a smarter reranker. Most RAG
quality problems are chunking problems.

## Method

1. **Split on structure before size.** Headings, sections, and
   paragraphs are meaning boundaries the author already provided, and
   respecting them beats any fixed character count.
2. **Size to the question, not to the model's limit.** Chunks large
   enough to contain a complete answer and small enough that most of the
   chunk is relevant, which is usually far below the context window.
3. **Overlap modestly at boundaries.** A small overlap prevents an
   answer that straddles a split from being lost, at the cost of some
   duplication in results.
4. **Carry context into the chunk.** Document title, section heading,
   and date prepended to the text, since a retrieved chunk arrives
   without its surroundings and must stand alone.
5. **Treat tables, code, and lists as units.** Splitting them produces
   fragments that are worse than useless because they look authoritative
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add amey-thakur/ai-skills/chunking-strategies@sha256:171f70916c6e3f55

The line at the top of this page pins the label git:20260728.fe84bd4, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_6apqon2lh2t2uhfp.svg)](https://markdownregistry.com/a/art_6apqon2lh2t2uhfp)

4 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 1 version of this file, committed on Jul 28, 2026.

versioncommittedcommitsizeaudit
git:20260728.fe84bd4 latest2026-07-28 fe84bd4 2,162 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (2162 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

amey-thakur/ai-skills · 14 stars · license MIT · pushed 2026-09-06 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_6apqon2lh2t2uhfp
GET https://markdownregistry.com/api/v1/resolve?ref=amey-thakur/ai-skills/chunking-strategies
GET https://markdownregistry.com/api/v1/blob/171f70916c6e3f55258e15bf8fd8adb0a8649c7f8cdbdd871ec68732c07a916f

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

More from amey-thakur/ai-skills

ai-skills Cursor rule
amey-thakur/ai-skills · .cursor/rules/ai-skills.mdc · Operate the AI-SKILLS library autonomously, selecting the right skill or prompt for each task.
git:20260721.9534dd9 · audit A · 14 stars
AGENTS.md AGENTS.md file
amey-thakur/ai-skills · AGENTS.md
git:20260728.aa354e7 · audit A · 14 stars
CLAUDE.md CLAUDE.md file
amey-thakur/ai-skills · CLAUDE.md
git:20260721.9534dd9 · audit A · 14 stars
llms.txt@docs llms.txt file
amey-thakur/ai-skills · docs/llms.txt
git:20260728.f55fde9 · audit B · 14 stars
llms.txt llms.txt file
amey-thakur/ai-skills · llms.txt
git:20260906.c8c8420 · audit B · 14 stars
accessibility-review SKILL.md skill
amey-thakur/ai-skills · skills/accessibility/accessibility-review/SKILL.md · Audit a UI for real accessibility, keyboard-first, with concrete failures and fixes rather than checkbox compliance…
git:20260718.7c1285c · audit A · 14 stars
accessible-forms SKILL.md skill
amey-thakur/ai-skills · skills/accessibility/accessible-forms/SKILL.md · Build forms with associated labels, identified errors, clear instructions, and autocomplete attributes. Use when…
git:20260728.fac4544 · audit A · 14 stars
alt-text-writing SKILL.md skill
amey-thakur/ai-skills · skills/accessibility/alt-text-writing/SKILL.md · Write alt text that conveys an image's purpose, marks decorative images empty, and handles functional and complex…
git:20260728.fac4544 · audit A · 14 stars
aria-usage SKILL.md skill
amey-thakur/ai-skills · skills/accessibility/aria-usage/SKILL.md · Use ARIA correctly by preferring native HTML, applying roles/states/properties only where needed, and following known…
git:20260728.fac4544 · audit A · 14 stars
color-contrast SKILL.md skill
amey-thakur/ai-skills · skills/accessibility/color-contrast/SKILL.md · Meet WCAG contrast ratios for text and non-text, never rely on color alone, and verify with tools including in dark…
git:20260728.fac4544 · audit A · 14 stars
focus-management SKILL.md skill
amey-thakur/ai-skills · skills/accessibility/focus-management/SKILL.md · Manage focus on route changes, in modals, and after actions, with visible focus indicators. Use when building…
git:20260728.fac4544 · audit A · 14 stars
keyboard-navigation SKILL.md skill
amey-thakur/ai-skills · skills/accessibility/keyboard-navigation/SKILL.md · Make interfaces fully keyboard-operable with logical focus order, no traps, skip links, and roving tabindex where…
git:20260728.fac4544 · audit A · 14 stars

Every file in amey-thakur/ai-skills

Browse by kind, by grade A, or by owner.