Process Inbox

Step 1: Read and classify each selected file

For each file, read the full content and determine the content type:

Show classification to the user before acting:

Classifying [filename]:
→ Type: [type]
→ Route to: [destination]
→ Summary: [2-sentence summary of the content]

Confirm? (Y / change type / skip)

Step 2: Process each confirmed file

For Meeting notes — delegate to the /meeting skill entirely:

• Pass the file path as input to the meeting skill's "specific file" mode
• The meeting skill handles everything: analysis, people cross-referencing, related context search, archiving (to Archive/Meetings Archive/), structured note creation, journal linking, and task extraction with user confirmation
• After the meeting skill completes, delete the original from Inbox/ (the meeting skill already archived it)
• Do NOT duplicate any meeting processing logic here — the meeting skill owns it

For all other types — process directly as a structured loose note:

• Location: Loose Notes/Work/YYYY-MM-DD - [Inferred Title].md
• Template: templates/loose-note.md
• Content: Organize the raw content — add a brief summary at the top, preserve original detail below
• Date: Use date from filename if present, otherwise today

For braindumps/journal fragments:

• If today's journal exists, append to ## Notes section as a linked note
• Create the loose note in Loose Notes/Work/ with full content

For tasks/action items:

• Preserve all - [ ] checkboxes in the note
• Include a "Source" line linking to the archive path

Step 3: Extract tasks from non-meeting content

This step applies to all non-meeting content types (decisions, ideas, references, task lists, braindumps). Meeting notes skip this step — the /meeting skill handles their task extraction.

Scan the processed content for potential tasks:

• Explicit action items: - [ ] checkboxes, "need to", "should", "follow up", "TODO"
• Commitments: "I'll", "I will", "I need to", "let me"
• Deadlines: anything with a date or timeframe attached to an action

ONLY extract tasks for the vault owner — skip tasks owned by others.

Check existing task files in Tasks/ to avoid duplicates (match by title similarity).

Present proposed tasks and wait for confirmation before creating:

📋 Potential tasks found in [filename]:
1. [Task description] — Priority: P3
2. [Task description] — Priority: P3

⏭️ Already exists (will skip):
- [[T-NNN - Existing task]]

Create these tasks? (Y/N or select which ones, e.g. "1" or "1,2")

After confirmation:

• Create task files in Tasks/ using the /task skill's create workflow (read counter, create files, increment counter)
• Source field: wikilink to the processed note [[YYYY-MM-DD - Title]]
• Status: inbox
• Regenerate Dashboard/Tasks.md
• Report created tasks: ✅ Created [[T-NNN - Title]]

If no potential tasks found, skip silently — don't show an empty tasks section.

Step 4: Archive the original (non-meeting files only)

Meeting files: already archived by the /meeting skill in Step 2 — skip.

All other files: after processing:

• Move original from Inbox/[filename] to Archive/Notes Archive/[filename]
• If a file with the same name exists in archive, append -2, -3, etc.

Step 5: Link in today's journal (non-meeting files only)

Meeting files: already linked by the /meeting skill — skip.

All other files: add a link in today's journal (journals/YYYY/MM-Month/YYYY-MM-DD.md) under ## Notes:

- [[YYYY-MM-DD - Title]] — processed from Inbox ([original filename])

If today's journal doesn't exist, skip this step.

Step 6: Summary

✅ Processed N file(s) from Inbox

[For each file:]
- [filename] → [[YYYY-MM-DD - Title]] ([type])
  [If tasks were created:] Tasks: [[T-NNN]], [[T-NNN]]

📁 Originals archived

📝 Linked in today's journal: [[YYYY-MM-DD]]

Inbox remaining: N files

Notes

• Skip README.md — never process or archive it
• If the file has a date in its name, use that as the note date — don't default to today unless no date is present
• If content type is unclear, always ask rather than guess
• For meeting notes found in Inbox, apply the full /meeting workflow — same quality, same output
• Preserve all original content — never discard anything from the raw file
• If a file is already structured (has frontmatter, has headers matching templates), warn the user before processing

Error Recovery

If no journal exists for today

• Skip journal linking, note it in the summary

If archive folder doesn't exist

• Create Archive/Notes Archive/ if needed before moving

If file cannot be classified

• Leave it in Inbox
• Show in summary as "skipped — needs manual classification"