sparse-reading · liuzedooong, sparseread-codex — Claude Code skill for file reading
The sparse-reading Claude Code skill examines large documents, structured files, and groups of project files. It chooses a reading route and gathers focused evidence.
MIT ★ 94 updated 2026-11-16
npx skills add Liuzedooong/SparseReading --skill sparse-reading -a claude-codeWhat sparse-reading does
sparse-reading routes file inspection through SparseRead tools. sro_decide suggests a route, while sro_preview returns a card and preview to help select evidence. Then sro_read takes a focused goal and a mode such as scout, focus, collect, or verify.
Use sro_raw for a narrow range or selector, with a raw_ref returned by SparseRead. sro_card retrieves file metadata; sro_trace explains recent routing. In Codex, bundled hooks stay inactive until the user reviews their definitions in /hooks and chooses to trust them.
What it helps with
- find relevant details in a large document
- inspect a structured file against a specific question
- gather evidence across project files
- trace why SparseRead chose a particular route
Example prompts for the agent
- Find the migration guidance in this long document and show the supporting passages.
- Inspect
config.jsonand collect the database settings. - Compare the project files that describe the authentication flow.
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/codex/plugin/sparseread-codex/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/codex/plugin/sparseread-codex/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 at ~/.claude/skills/sparse-reading/SKILL.md.
What does sparse-reading do?
It routes inspection of large documents, structured files, and collections of project files through SparseRead tools. Those tools help choose a reading method and retrieve focused evidence.
Which SparseRead tools does sparse-reading use?
sro_decide and sro_preview help select a route; sro_read reads with a chosen mode. The skill also uses sro_raw, sro_card, and sro_trace.
Source text
SKILL.md · MIT
--- name: sparse-reading description: Use SparseRead tools to inspect large, structured, or multi-file project inputs with targeted evidence. --- # SparseRead in Codex Use SparseRead when an answer depends on a large document, structured file, or a collection of project files. - Call `sro_decide` to inspect the recommended reading route for a path. - Call `sro_preview` before exploring a large or structured input. Use the returned card and preview to choose the evidence needed. - Call `sro_read` with a focused goal and suitable mode such as `scout`, `focus`, `collect`, or `verify`. - Call `sro_raw` only with a `raw_ref` returned by SparseRead and a narrow range or selector. - Call `sro_card` for file metadata and `sro_trace` to explain recent SparseRead routing. Small and bounded reads can use Codex's normal tools. If a hook redirects a broad shell read, preview the named path and use the evidence tools before retrying the native command. Codex does not run the bundled hooks until you review and trust their definitions in `/hooks`. Trust is a user decision; the plugin does not bypass it.
© Liuzedooong · MIT · Liuzedooong/SparseReading
Similar skills
All skills for Agent setup: memory, context, orchestration →sparse-reading by other authors comes first.