Readwise Highlight Search Skill

Readwise Highlight Search

Decision Tree

1. Do you know the EXACT document or author?
   YES -> Use fulltext filters (--author, --title, --text)
   NO  -> Use vector search (semantic query)
2. Need document-level results (not highlights)?
   YES -> Use `readwise list` with --tag or --category filters
   NO  -> Use `readwise search`

Quick Reference

Vector + Fulltext Search (Highlights)

# Semantic search
readwise search "fiduciary duty broker-dealer"

# With filters
readwise search "regulation" --author "Jackson" --title "Financial"
readwise search "proxy" --highlight-tag "corps"

# JSON output for piping
readwise search "query" --json

Fulltext filter fields:

| Flag | Searches | |------|----------| | --author | Document author name | | --title | Document title | | --note | Highlight notes/annotations | | --text | Highlight text content | | --highlight-tag | Tags on highlights |

Document Search (Reader API)

# By tag
readwise list --tag "proxy advisors" --json

# By category
readwise list --category pdf --location archive

# Recent updates
readwise list --updated-after 2026-01-01T00:00:00Z --limit 20

Highlight Keyword Search (v2 API)

# Simple text search across highlights
readwise highlights --search "fiduciary" --limit 20 --json

When to Use Each

| Need | Command | Speed | |------|---------|-------| | Semantic/conceptual search | readwise search "query" | ~2s | | Keyword in highlight text | readwise highlights --search "term" | ~1s per page | | Documents by tag | readwise list --tag "X" | ~1s per page | | Specific document | readwise get <id> | ~0.5s |

Output Format

Use --json for structured output. Without it, results are printed as aligned tables.

JSON search results include a score field (higher = more relevant, >0.01 typically meaningful).

Agent Skills: Readwise Highlight Search

Install this agent skill to your local

Skill Files