Add to Reading Inbox

Triage reading material into the Obsidian vault inbox (synced via Syncthing) and optionally to Readwise Reader.

Inputs

Workflow

User says "add X to reading list"
    |
    v
Is the source already a file path or URL?
    |
  YES --> Go to Step 2 (metadata extraction)
    |
   NO --> Step 1: Resolve the source
          |
          User said "attachment" / "inbox" / "email"?
            YES --> superhuman search → download attachment → proceed
          User gave a paper title or author?
            YES --> Search email FIRST, then scholar/web only if not in email
          User pointed to a calendar event?
            YES --> Check event for attachments/links → proceed
          User mentioned "Canvas" / "posted on Canvas"?
            YES --> Canvas API: list modules → find file → download → proceed

Steps

1. Resolve Source (when source is not yet a file or URL)

The user often says things like "find the paper from the workshop and add it to reading" without providing a direct path. Your job is to locate the actual file.

Source resolution priority — follow this order:

Downloading email attachments:

# Search for the email
superhuman search "workshop paper" --limit 5

# Download the attachment (use the message ID from search results)
superhuman attachment download <message-id> --output /tmp/

After downloading, proceed to Step 2 with the downloaded file path.

Downloading from Canvas LMS:

If the user mentions Canvas, use the Canvas API to find and download the file. See references/canvas-courses.md for course IDs, credentials location, and API patterns.

2. Extract Metadata & Rename (files only)

Use look-at to extract metadata from the first page:

LOOK_AT=$(${CLAUDE_SKILL_DIR}/../../skills/look-at/scripts/look_at.py) && python3 "$LOOK_AT" \
    --file "/path/to/file.pdf" \
    --goal "Extract the title, authors, and date/year from the first page. Return as: Title: ..., Authors: ..., Date: ..."

Naming convention: Author (Date) - Title.ext

Use last names only. Truncate long titles to a readable short form.

Confirm the proposed filename with the user before copying. Gemini may misread abbreviated dates (e.g., 25-Feb-26 → wrong year). Always verify.

3. Copy to Vault Inbox

mkdir -p "/Users/vwh7mb/Documents/Notes/Vault/3. Resources/Inbox"
cp "/path/to/source.pdf" "/Users/vwh7mb/Documents/Notes/Vault/3. Resources/Inbox/Author (Date) - Title.pdf"

The vault syncs via Syncthing — file is accessible on all devices (Onyx Boox, etc.) without any extra upload step.

Multiple attachments: If the source email has multiple related files (e.g., paper + appendices), copy all of them. Name appendices consistently: Author (Date) - Title - Appendix.pdf.

4. Add to Daily Note (under Reading heading)

The daily note has multiple sections. Insert the entry under the # Reading heading, not at the end of the file.

Step 4a: Ensure daily note exists

obsidian vault=Vault daily:read >/dev/null 2>&1 || obsidian vault=Vault daily:append content=""

If obsidian CLI is unavailable, create manually:

TODAY=$(date +%Y-%m-%d)
DAILY="/Users/vwh7mb/Documents/Notes/Vault/3. Resources/Daily Notes/${TODAY}.md"
if [ ! -f "$DAILY" ]; then
  cat > "$DAILY" << 'TMPL'
# To-Dos
# Reading
# Meetings
# Work
TMPL
fi

Step 4b: Insert under Reading heading

Use the Edit tool to insert the new line after # Reading (or after the last existing reading item under that heading).

Format rules:

• Always a checkbox: - [ ]
• For files: use an Obsidian wiki-link to the renamed file: [[Author (Date) - Title.pdf]]
• Project backlink inline if provided: [[Author (Date) - Title.pdf]] [[Project Name]]
• No backlink if no project context: [[Author (Date) - Title.pdf]]
• For URLs: use a markdown link: [Title](https://example.com)
• Never add task metadata (dates, priority). The Obsidian Tasks plugin handles that automatically.

5. Save to Readwise Reader (optional)

Only when user explicitly requests Readwise.

For URLs:

readwise reader-create-document --url "https://example.com/article" --tags reading-inbox

For files (PDFs, EPUB, etc.):

readwise upload "/path/to/file.pdf" --tag reading-inbox

6. Confirm

Report what was done:

• Renamed filename
• Where the file was saved
• What was added to the daily note
• Whether it was saved to Readwise

Examples

"Add this PDF to my reading list" (file path provided)

1. look-at extracts: Title: Mirroring the Market, Authors: Nathan Atkinson, Date: 2026
2. Propose: "Atkinson (2026) - Mirroring the Market.pdf" → user confirms
3. cp to vault inbox
4. Daily note: "- [ ] [[Atkinson (2026) - Mirroring the Market.pdf]]"

"Find the paper attachment for the workshop and put it on my reading list" (source not provided)

1. superhuman search "workshop paper" → find email with PDF attachment
2. superhuman attachment download → /tmp/paper.pdf
3. look-at extracts: Title: Canons and the Court, Authors: Jonathan Choi, Nina Mendelson, Date: 2026
4. Propose: "Choi, Mendelson (2026) - Canons and the Court.pdf" → user confirms
5. cp to vault inbox
6. Daily note: "- [ ] [[Choi, Mendelson (2026) - Canons and the Court.pdf]]"

"Add this paper to reading, related to Pass-through Voting"

Same as above, daily note entry:
"- [ ] [[Atkinson (2026) - Mirroring the Market.pdf]] [[Pass-through Voting]]"

"Save this article to reading and readwise"

1. Daily note: "- [ ] [Article Title](https://example.com/article)"
2. readwise reader-create-document --url "https://example.com/article" --tags reading-inbox

Red Flags — STOP If You Catch Yourself:

Iron Law: Correct Section Placement

NEVER use daily:append for reading items. It appends to the end of the note (after Work section). Instead, always Read the daily note file and Edit to insert under the # Reading heading.

WRONG: obsidian vault=Vault daily:append content="- [ ] Paper"
  → Ends up after # Work

RIGHT: Read daily note → Edit to insert after # Reading line
  → Entry appears in correct section

Iron Law: Attachment Means Email

When the user says "attachment", "the paper someone sent", or "that PDF from [person/event]" — search email FIRST. Do not search the web, SSRN, Google Scholar, or any other source before checking email.

The user's mental model: someone sent them a file → it's in their inbox. Searching elsewhere first is a detour that wastes time, burns context, and forces the user to repeat themselves.

WRONG: User says "attachment" → search SSRN → search web → search scholar → finally check email
  → 12+ tool calls wasted, user frustrated

RIGHT: User says "attachment" → superhuman search → download → done
  → 2-3 tool calls, user happy