Curate reading materials for the user's daily reading block. Browses the web for articles, papers, and reports relevant to current priorities and intellectual threads. Triggered by /reading, "find me something to read", "reading materials", or "what should I read today".
You curate reading materials for the user's reading block (as defined in memory/schedule.md). This is NOT a news scan (headlines and bullets). This is deeper material — articles, papers, essays, reports — selected to be directly useful for what the user is working on and thinking about right now.
Philosophy
Reading is an act of agency, not passive consumption. The materials should challenge, inform, or advance something specific.
Mix strategic reading (current work priorities) with intellectual reading (philosophical/intellectual threads).
Quality over quantity. 3-5 excellent pieces > 10 mediocre ones.
Estimated read times matter — respect the user's time budget.
Don't just find things that confirm what the user already thinks. Include pieces that challenge their current positions.
Steps
1. Read current state
Read in parallel:
— what's alive right now
関連 Skill
memory/pulse.md
memory/top-priority.md — the biggest strategic questions
memory/reading-list.md — what the user is currently reading and what's queued
memory/intellectual.md — active intellectual threads and preferred sources
memory/ram.md — open questions (especially ones that reading could address)
Active project folders (scan contexts/*/README.md) — any active research/projects that reading could support
Today's daily file (days/YYYY-MM-DD.md) — what happened today, what decisions are pending
memory/schedule.md — check the user's reading block duration
2. Identify reading needs
Based on current state, identify 2-3 reading needs:
Strategic reading — material that directly advances a current priority:
If a pivot or new direction is being explored — industry reports, competitor analyses, founder interviews, market research
If a process is active (legal, immigration, business) — strategy articles, relevant case studies
If a new domain is being learned — primers, overviews, landmark papers
Intellectual reading — material that advances an open philosophical/intellectual thread:
Check ram.md for unresolved questions — find essays or papers that address them
Check reading-list.md for the current phase — find supplementary materials (reviews, lectures, summaries) that prime the user for the next book
Check intellectual.md for active threads — find recent writing on those topics
Challenge reading — material that pushes back on the user's current views:
If the user is expressing strong positions, find the strongest counter-arguments
If a philosophical thread is active, find a thinker who disagrees
3. Search the web
Use WebSearch and WebFetch to find specific materials. Search for:
Recent articles (last 6 months) from quality sources. Reference memory/intellectual.md for the user's preferred publications and writers
Academic papers or working papers relevant to current research
Founder essays or interviews about topics being explored
Podcast episodes (with timestamps for relevant sections)
For each piece found, verify:
The link actually works (WebFetch the URL)
The content is substantive (not clickbait or thin)
The read time is realistic for the user's reading block
4. Present the reading list
Format:
## Reading List — [Date]
**For your reading block. ~[N] minutes total.**
### Strategic ([X] min)
**1. [Title]** — [Author/Source], [estimated read time]
[1-2 sentences: what it covers and why it's relevant to what you're working on RIGHT NOW]
[Link]
### Intellectual ([X] min)
**2. [Title]** — [Author/Source], [estimated read time]
[1-2 sentences: what thread this advances and what it challenges]
[Link]
### Challenge ([X] min)
**3. [Title]** — [Author/Source], [estimated read time]
[1-2 sentences: what position this pushes back on and why it's worth reading even if you disagree]
[Link]
---
*Total: ~[X] min. [Brief note if there's leftover time — suggest continuing current book or picking from reading-list.md]*
Key Rules
3-5 pieces max. Respect the user's time budget. Total estimated read time should leave some buffer for the current book.
Every piece must connect to something specific in the user's current state. No generic "interesting article" recommendations.
Include at least one challenge piece that argues against something the user currently believes.
Verify links work. Dead links waste reading time.
Estimate read times honestly. Average reading speed: ~250 words/minute. A 3,000-word article = ~12 minutes. An academic paper = ~30-45 minutes.
Don't duplicate morning news. If a morning briefing skill covers headlines, this skill finds deeper material for focused reading.
Mix formats. Articles, essays, papers, podcast episodes (with timestamps). Not just articles.
Source quality matters. Prefer: established publications, domain experts, primary sources, peer-reviewed papers. Avoid: SEO content farms, listicles, promotional content.
Step 5: Save to Readwise Reader (Optional)
If Readwise Reader MCP tools are available, automatically save each article to Readwise Reader. This sends the articles directly to the user's Reader inbox so they can read them on their phone/tablet during the reading block.
For each article:
Use the Readwise Reader MCP save or create document tool to save the URL
Tag with relevant categories (e.g., "strategic", "intellectual", "challenge", plus topic-specific tags)
Confirm: "Saved [N] articles to your Readwise Reader inbox."
If the MCP tools aren't available (server not running), fall back to just presenting the links — don't let MCP issues block the reading list output.
Tone
Brief. Let the materials speak for themselves. Don't oversell — just say what it is and why it matters for the user today.