readwise-reader（明智的读者）

Readwise Reader Skill

Use this skill to script workflows against the Readwise Reader product (saved articles, feeds, PDFs, newsletters).

Quick start

1. Generate a token from https://readwise.io/access_token and store it in READWISE_TOKEN.
2. Call uv run --project ${CLAUDE_PLUGIN_ROOT} python ${CLAUDE_PLUGIN_ROOT}/skills/readwise-reader/scripts/reader_client.py... whenever possible; it handles retries, .generated tagging, and --dry-run against the /api/v3 endpoints.
3. Respect Reader's tighter rate limits (20 req/min). The CLI surfaces remaining budget whenever headers are present; throttle accordingly.

Example workflows

• "Save this article to Reader for later" — uses docs create with a URL
• "Show me what's new in my Reader inbox" — uses docs list --category new
• "Archive everything I've finished reading this week" — uses docs pull + docs update --state archive

CLI commands

• uv run --project ${CLAUDE_PLUGIN_ROOT} python ${CLAUDE_PLUGIN_ROOT}/skills/readwise-reader/scripts/reader_client.py docs create --url <article> [--title... --summary... --location later --tags... --labels... --generated --dry-run] – creates a document via URL or raw HTML (--content). Use --location to set where it lands (defaults to later). Reader API v3 does not support uploading local files directly.
• ... docs list [--location later] [--category article] [--tag deep-work] [--limit 25] [--id <doc-id>] [--with-content] – paginated document listing. Defaults to --location later. Use --id to fetch a single document. Use --with-content to include html_content (article text, video transcripts).
• ... docs update <id> [--state archive --tags "deep,focus" --title...] – patch metadata/state (new, later, archive). Supports --dry-run.
• ... docs pull --since 2026-02-01 – fetches documents updated since a timestamp for recap/triage workflows.
• ... auth validate – confirms your token works by hitting the /api/v2/auth/ endpoint.

Locations vs categories

Locations describe where a document sits in your reading workflow:

• new — inbox, freshly added items awaiting triage
• later — saved for later reading (the default for docs list)
• shortlist — prioritized for near-term reading
• archive — finished or dismissed
• feed — RSS/feed items not yet triaged

Categories describe the content type:

• article, email, rss, highlight, note, pdf, epub, tweet, video

When a user asks for their "inbox" or "reading list", query --location later (the default). Use --location new only when specifically checking for untriaged items.

Default output is human-readable markdown with only key fields. Use --raw to get full JSON with all fields.

Fetching content & transcripts

Use --with-content to retrieve the full html_content field, which contains:

• Articles: full article text as HTML
• Videos: auto-generated transcript (for YouTube videos with captions)
• PDFs/EPUBs: extracted text content

This is disabled by default to keep responses fast. When you need to analyze, summarize, or extract quotes from a document, fetch it by ID with content:

... docs list --id <doc-id> --with-content --raw

Data guidance

• Generated entries: set --generated when saving synthetic journal entries or agent summaries. The CLI appends .generated to tags (and labels, when provided) so they are searchable in Reader.
• Source inputs: prefer --url when the content exists online. Use --content for local snippets; Reader API v3 does not expose the legacy upload flow, so convert PDFs to shareable URLs before saving.
• Metadata: Reader tolerates arbitrary tags/labels, so lean on them for downstream automations (e.g., .journal, .deepread). Avoid overloading summary with custom formats—store structured metadata in labels/tags instead.
• Dry runs: --dry-run prints the payload without hitting the API. Dry-run output always shows the full payload (no field filtering).

Scripts

• ${CLAUDE_PLUGIN_ROOT}/skills/readwise-reader/scripts/reader_client.py: CLI covering document create/list/update/pull plus token validation against Reader API v3. Integrates with shared utilities from ${CLAUDE_PLUGIN_ROOT}/readwise_common/.
• Shared helpers live in ${CLAUDE_PLUGIN_ROOT}/readwise_common/ (auth, HTTP retries, tag/location utilities); import from there when extending functionality to keep behavior consistent.

Testing & validation

• Use uv run --project ${CLAUDE_PLUGIN_ROOT} python -m compileall ${CLAUDE_PLUGIN_ROOT}/skills/readwise-reader ${CLAUDE_PLUGIN_ROOT}/readwise_common before shipping changes.
• Smoke-test live calls (token required) with docs create --dry-run, docs list --limit 5, and docs pull --since <yesterday> to confirm pagination + tagging rules.