sparse-reading · liuzedooong, openclaw — Claude Code skill: what it does, how to install
The sparse-reading Claude Code skill extracts targeted evidence from long documents, PDFs, and compact file collections. Small files stay on the usual read path.
MIT ★ 94 updated 2026-11-16
npx skills add Liuzedooong/SparseReading --skill sparse-reading -a claude-codeWhat sparse-reading does
The skill previews a large file, PDF, report, or evidence directory with sro_preview before doing a broader read. When specific evidence is needed, sro_read can collect answers for several questions using explicit slots. Once the result is ready, it moves on to the requested report or other deliverable.
Small files, configuration tasks, exact table calculations, and script-driven analysis stay with native read, grep, and exec workflows. sro_raw is for cases that need the original content. In OpenClaw’s native profile, SparseRead is not pushed; advisory leaves native tools available, while enforce routes broad reads and searches through sro_preview.
For Claude Code, installation creates ~/.claude/skills/sparse-reading/SKILL.md.
What it helps with
- inspect a long document or PDF before reading it broadly
- collect answers to several questions across a file set
- locate supporting evidence in a compact directory
- keep small configuration files on native read tools
Example prompts for the agent
- Find the dates and terms in this large PDF.
- Collect answers to these questions from the files in
evidence. - Gather supporting evidence for a report from the attached documents.
How to install sparse-reading
1Skills CLI (Vercel)
npx skills add Liuzedooong/SparseReading --skill sparse-reading -a claude-code2Manually: put the files in place
~/.claude/skills/sparse-reading/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/sparse-reading/
1Skills CLI (Vercel)
npx skills add Liuzedooong/SparseReading --skill sparse-reading -a codex2Manually: put the files in place
.agents/skills/sparse-reading/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add Liuzedooong/SparseReading --skill sparse-reading -a cursor2Manually: put the files in place
.cursor/skills/sparse-reading/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install Liuzedooong/SparseReading integrations/openclaw/plugin/skills/sparse-reading2Manually: put the files in place
.github/skills/sparse-reading/SKILL.md1Skills CLI (Vercel)
npx skills add Liuzedooong/SparseReading --skill sparse-reading -a gemini-cli2Platform CLI
gemini skills install https://github.com/Liuzedooong/SparseReading --path integrations/openclaw/plugin/skills/sparse-readingManually: put the files in place
.devin/skills/sparse-reading/SKILL.mdThe legacy .windsurf/skills/ path still works
Claude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about sparse-reading
How do I install sparse-reading in Claude Code?
Run npx skills add Liuzedooong/SparseReading --skill sparse-reading -a claude-code. The skill file is placed at ~/.claude/skills/sparse-reading/SKILL.md.
What does sparse-reading do?
It previews long documents, PDFs, and compact file collections, then gathers specific evidence for the requested task. Small files remain on the native reading path.
When does sparse-reading use native read tools?
Native reads remain available for small files, setup files, templates, unsupported objects, and named unresolved files. The OpenClaw native profile does not push SparseRead.
Source text
SKILL.md · MIT
---
name: sparse-reading
description: Use SparseRead for long documents, PDFs, and compact multi-file evidence closures without disrupting small native reads.
---
Use this skill only when a tool result, `sro_preview`, or the task shape indicates
SparseRead is useful. Keep ordinary native `read`, `grep`, and `exec` workflows
for small files, small config/code/data tasks, exact full-table computation, and
script-driven analysis.
SparseRead protocol:
1. Call `sro_preview(path)` on a large file, PDF/report, or compact evidence
directory before broad native reading. Use the preview directly when enough.
2. Continue with `sro_read({ artifact_id }, mode, hint)` only when targeted
evidence is needed. For multi-question documents and evidence closures,
prefer `mode: "collect"` with explicit `slots`.
3. If `slot_digest.overall_status` is `ready` or `next_action.allowed_next`
allows writing, write the requested deliverable. Do not keep broad-reading
or repeating `sro_read` after ready.
4. Native small reads are still allowed for setup files, templates, unsupported
objects, and named unresolved slots.
5. Use `sro_raw(raw_ref)` only when the original content is explicitly needed.
`sro_card -> sro_read` remains for compatibility and `bench_protocol`, not as
the normal production path.
OpenClaw gate profiles:
- `native`: do not push SparseRead.
- `advisory`: SparseRead may help, but native tools stay available.
- `enforce`: broad native read/search/exec-dump should use
`sro_preview` first.
Command-security closures are advisory, not hard replacement. Use one
`sro_read(mode="collect")`; once ready, write the report/JSON immediately. Read
small templates or named unresolved files natively when needed.
© Liuzedooong · MIT · Liuzedooong/SparseReading
Similar skills
SKILL.md skills catalog →sparse-reading by other authors comes first.