Log Triage - Daily Note Processing

Process daily note log entries into structured notes. Extract voice memos into meeting notes, URLs into clipping notes, and reminders into capture tasks.

Templates

Use para_config to get the vault and templates path. Use para_template_fields to inspect required fields for any template.

Primary templates: meeting, capture Clipping templates: In Templates/Clippings/ - matched by URL domain (see URL mapping below)

The Flow

Log entry → LLM extraction → Suggest note type → User confirms → Create note → Delete log entry

Log Entry Types

1. Voice Memos (🎤)

Pattern: - [time] - 🎤 [transcription]

→ Extract into meeting note format, create with type: meeting

2. URLs

Pattern: - [time] - https://...

→ Always try Firecrawl first → Detect clipping type → Create clipping note

If Firecrawl fails (auth, blocked, timeout) → Still create clipping note by parsing the URL.

URL → Clipping Type Mapping:

Domain/Path Pattern	Clipping Template	Extract From URL
youtube.com/watch?v=	youtube-video.md	Video ID, title from path
github.com/{owner}/{repo}	github-repo.md	Owner, repo name
*.atlassian.net/wiki/	documentation.md	Space, page title
docs.*, developer.*	documentation.md	Path segments
medium.com, substack.com, news sites	article.md	Title from path
twitter.com, x.com	tweet---x-post.md	Tweet ID
reddit.com/r/{sub}/comments/	reddit-post.md	Subreddit, post title
stackoverflow.com/questions/	stack-overflow.md	Question ID, title
en.wikipedia.org/wiki/	wikipedia.md	Article title
goodreads.com/book/	book.md	Book title
imdb.com/title/	movie.md	Movie title
spotify.com/episode/, podcast sites	podcast-episode.md	Episode title
udemy.com, coursera.org	course---tutorial.md	Course title
amazon.com, shopping sites	product---gift-idea.md	Product title
airbnb.com, booking.com	accommodation.md	Listing title
yelp.com, restaurant sites	restaurant.md	Restaurant name
eventbrite.com, meetup.com	event.md	Event title
maps.google.com, goo.gl/maps	place.md	Place name
chatgpt.com/share/	chatgpt-conversation.md	Conversation ID
claude.ai/share/	claude-conversation.md	Conversation ID
apps.apple.com, play.google.com	app---software.md	App name
Default	article.md	Best guess from path

3. Text Entries / Short Voice Memos

Pattern: - [time] - [text] or - [time] - 🎤 [short text]

Reminder/task signals → Create capture note with Obsidian Tasks checkbox inside:

• "remind me to..."
• "don't forget..."
• "remember to..."
• "I need to..."
• "tomorrow I should..."

Thought/idea → Create capture note with source: thought

Noise → Just delete

Session Flow

1. Read daily note - Load 000 Timestamps/Daily Notes/YYYY-MM-DD.md
2. Parse log section - Find entries under ## Log
3. Suggest note types first - Show user what each entry looks like:

   Found 3 log entries:
   
   1. 🎤 9:26am (~2000 words) → Suggest: Meeting note
   2. 🎤 12:04pm (~3000 words) → Suggest: Meeting note
   3. URL 1:33pm (atlassian.net) → Suggest: Clipping (documentation, auth required)
   
   Proceed with these? [y/n/change]
   

4. For each confirmed entry:
   • Extract structure into note format
   • Show extraction preview
   • Ask user to confirm
   • Create note in 00 Inbox
   • Delete entry from daily note
5. Summary - What was created, what was skipped

Voice Memo → Meeting Note

Extract the transcription into meeting note format:

---
type: meeting
meeting_type: general
meeting_date: 2026-01-06
area: "[[Career & Contracting]]"
company: Bunnings
template_version: 1
---

# GMS Project Kickoff - Jackie

## Attendees

- Jackie (Team Lead, on leave until 27th)
- Josh Green (Tech Lead, Perth)
- June (Frontend/Backend dev, my buddy)

## Notes

- GMS = Gift Card Management System
- Reseller integration with Black Hawk
- Deadline: July (likely to extend)

## Decisions Made

- Estimates in days, not story points
- All-in day: Thursday

## Action Items

- [ ] Get MacBook bun number to Jackie
- [ ] Test VPN from home tonight
- [ ] Look at Miro board for resellers

Key sections to extract:

• Attendees - Names and roles mentioned
• Notes - Key information, context, project details
• Decisions Made - Any conclusions or agreements
• Action Items - Tasks as checkboxes (these show up in Dataview!)

Raw Transcription Section (CRITICAL)

The meeting template does NOT have a Raw Transcription section. You MUST add it after creating the note.

Use para_insert with mode: "after" on the ## Follow-up heading to append:

---

## Raw Transcription

> [Full cleaned transcription in blockquote format, broken into paragraphs]

Transcription Cleanup Rules

Before inserting, apply basic cleanup to the raw transcription:

• Remove filler words: "um", "uh", "like" (when filler), "you know", "sort of"
• Fix grammar: Correct obvious errors, add missing words
• Add punctuation: Proper periods, commas, question marks
• Break into paragraphs: At natural topic changes or speaker switches
• Preserve tone: Keep conversational style and ALL content

DO NOT summarize or truncate. The entire transcription must be preserved.

Reference example: 02 Areas/🤝🏻 Contract - Bunnings/🗣️ IT Onboarding - Bunnings.md

URL → Clipping Note

For URLs, fetch content and create appropriate clipping:

1. Fetch URL with firecrawl_scrape
2. Detect clipping type from domain/content:
   • youtube.com → youtube-video
   • github.com → github-repo
   • twitter.com, x.com → tweet (see X/Twitter section below)
   • *.atlassian.net → documentation (may need auth)
   • News/blog sites → article
   • etc.
3. Extract fields from page content:
   • Title, author, published date, domain
   • Summary (first paragraph or meta description)
4. Create clipping note using template structure

X/Twitter Posts (Special Handling)

For twitter.com or x.com URLs, create notes matching the WebClipper output.

⚠️ X requires JavaScript - Firecrawl and WebFetch cannot scrape X. Use one of:

1. Chrome DevTools MCP (mcp__chrome-devtools__*) - Navigate and extract via DOM
2. Manual entry - Ask user for author/content if Chrome DevTools unavailable
3. URL parsing - Extract username and tweet ID from URL as fallback

Chrome DevTools approach:

1. mcp__chrome-devtools__navigate_page({ url: "https://x.com/..." })
2. mcp__chrome-devtools__take_snapshot() - Get page content
3. mcp__chrome-devtools__evaluate_script() - Extract specific fields

Extract from page:

• Author: Display name (e.g., "Matt Pocock")
• Handle: @username (e.g., "@mattpocockuk")
• Posted: Date from the post
• Content: Full post text (tweets or X Articles)

AI generation (use context from scraped content):

• ai_summary: "In 10 words or less, what is the key point of this post?"
• topics: "List 2-3 topic tags as comma-separated list (e.g. tech, AI, programming)"
• sentiment: "Is this post positive, negative, neutral, or controversial? One word."

Template output:

---
type: clipping
clipping_type: tweet
source: "https://x.com/mattpocockuk/status/123456789"
clipped: 2026-01-09
author: Matt Pocock
handle: "@mattpocockuk"
posted: 2026-01-09
ai_summary: AI coding agent loops using Ralph and Claude
topics: AI, programming, automation
sentiment: positive
distill_status: raw
related: []
project: []
area: []
---

# ✂️🐦 Matt Pocock - AI Coding Agents With Ralph

## AI Summary

> This post introduces Ralph, a technique for running AI coding agents iteratively...

---

**Author:** Matt Pocock (@mattpocockuk)
**Posted:** 2026-01-09
**Clipped:** 2026-01-09
**Link:** [View on X](https://x.com/mattpocockuk/status/123456789)

## Content

[Full post content from Firecrawl scrape]

## My Notes

Note naming: ✂️🐦 {Author} - {AI-generated 3-6 word topic summary}

Generate a unique title based on the tweet's content (e.g., "AI Coding Agents With Ralph", "TypeScript Inference Tips"). This prevents filename collisions when the same author posts multiple tweets.

YouTube Videos (Special Handling)

For youtube.com or youtu.be URLs, use the YouTube MCP tools to fetch video info and transcript.

Step 1: Fetch video info and transcript

mcp__youtube-transcript__get_video_info({ url: "https://www.youtube.com/watch?v=..." })
mcp__youtube-transcript__get_transcript({ url: "https://www.youtube.com/watch?v=..." })

Extracted fields from video info:

• title: Video title
• uploader: Channel name
• description: Video description
• upload_date: Published date (format: YYYY-MM-DDTHH:MM:SS)
• duration: Video length (e.g., "4 minutes")

Step 2: Create note using youtube-video template

para_create({
  template: "youtube-video",
  title: "✂️🎬 {Video Title}",
  dest: "00 Inbox",
  args: {},
  content: {
    "Description": "[Video description from API]"
  },
  response_format: "json"
})

Step 3: Set frontmatter with extracted values

para_frontmatter_set({
  file: "00 Inbox/✂️🎬 {Video Title}.md",
  set: {
    source: "https://www.youtube.com/watch?v=...",
    video_id: "dQw4w9WgXcQ",
    channel: "Rick Astley",
    duration: "4 minutes",
    published: "2009-10-25",
    transcript_status: "complete"
  }
})

Step 4: Insert transcript after the placeholder

The template has <!-- transcript:pending -->. Replace it with the full transcript:

para_insert({
  file: "00 Inbox/✂️🎬 {Video Title}.md",
  heading: "Transcript",
  content: "[Full transcript from API]",
  mode: "append"
})

Step 5: Generate AI Summary

Based on the description and transcript, generate a 3-bullet summary:

para_insert({
  file: "00 Inbox/✂️🎬 {Video Title}.md",
  heading: "AI Summary",
  content: "> - Key insight 1\n> - Key insight 2\n> - Key insight 3",
  mode: "append"
})

Template output:

---
type: clipping
clipping_type: youtube-video
source: "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
video_id: dQw4w9WgXcQ
channel: Rick Astley
duration: 4 minutes
published: 2009-10-25
clipped: 2026-01-09
consumption_status: to-watch
transcript_status: complete
distill_status: raw
related: []
project: []
area: []
---

# ✂️🎬 Rick Astley - Never Gonna Give You Up (Official Video)

## AI Summary

> - Classic 1987 hit that topped charts in 25 countries
> - Directed by Simon West (later directed Con Air, Tomb Raider)
> - Passed 1 billion YouTube views in July 2021

---

![](https://i.ytimg.com/vi/dQw4w9WgXcQ/maxresdefault.jpg)

**Channel:** Rick Astley
**Duration:** 4 minutes
**Published:** 2009-10-25

## Description

The official video for "Never Gonna Give You Up" by Rick Astley...

## Notes



## Transcript

♪ We're no strangers to love ♪
♪ You know the rules and so do I ♪
...

Note naming: ✂️🎬 {Channel} - {Video Title} (truncate to 80 chars)

---
type: article
clipping_type: article
source: "https://example.com/article"
clipped: 2026-01-06
author: "John Smith"
published: "2026-01-05"
domain: "example.com"
distill_status: raw
related: []
project: []
area: []
template_version: 1
---

# Article Title

## AI Summary

- Key point 1
- Key point 2
- Key point 3

---

**Author:** John Smith
**Published:** 2026-01-05
**Source:** [example.com](https://example.com/article)

---

## Summary

Brief description of the article...

## Highlights



## Full Content

[First 15000 chars of content]

For YouTube videos: Also fetch transcript with mcp__youtube-transcript__get_transcript

If Firecrawl fails: Parse URL and create clipping anyway:

URL: https://bunnings.atlassian.net/wiki/spaces/POS/pages/15205007376/Onboarding+Plan+for+Nathan+Vale

• Domain: bunnings.atlassian.net → type: documentation
• Path contains /wiki/spaces/POS/ → Confluence, space: POS
• Last segment: Onboarding+Plan+for+Nathan+Vale → title: "Onboarding Plan for Nathan Vale"

---
type: documentation
clipping_type: docs
source: "https://bunnings.atlassian.net/wiki/spaces/POS/pages/15205007376/Onboarding+Plan+for+Nathan+Vale"
clipped: 2026-01-06
domain: "bunnings.atlassian.net"
distill_status: raw
related: []
project: []
area: "[[Career & Contracting]]"
template_version: 1
---

# Onboarding Plan for Nathan Vale

## AI Summary

Confluence documentation page (POS space). Content requires authentication.

---

**Source:** [bunnings.atlassian.net](https://bunnings.atlassian.net/wiki/spaces/POS/pages/15205007376/Onboarding+Plan+for+Nathan+Vale)

---

## Why I Saved This

Text Entry / Short Voice Memo → Capture Note

For text entries or short voice memos (reminders, thoughts, ideas):

Reminder example ("remind me to...", "don't forget..."):

---
type: capture
status: inbox
source: reminder
resonance: useful
urgency: high
template_version: 1
---

# Take out the trash

## Capture

- [ ] Take out the trash 🔔 2026-01-06 22:00

## Why I Saved This

From daily log on [[2026-01-06]]

Task emoji format (Obsidian Tasks plugin):

• 📅 = due date
• 🔔 = reminder/scheduled
• ⏳ = scheduled date
• 🔁 = recurring

Thought example:

---
type: capture
status: inbox
source: thought
resonance: inspiring
urgency: low
template_version: 1
---

# Tomorrow is a new day

## Capture

Don't forget tomorrow is a new day.

## Why I Saved This

From daily log on [[2026-01-06]]

Example Triage Session

Found 3 log entries in 2026-01-06:

1. 🎤 9:26am (~2000 words)
   Looks like: IT onboarding meeting
   → Create: Meeting note "IT Onboarding - Bunnings"
   → Tasks: 4 action items extracted

   Create this? [y/n]

2. 🎤 12:04pm (~3000 words)
   Looks like: Project kickoff with Jackie
   → Create: Meeting note "GMS Project Kickoff - Jackie"
   → Tasks: 5 action items extracted

   Create this? [y/n]

3. URL 1:33pm - Confluence link (bunnings.atlassian.net)
   → Create: Clipping "Onboarding Plan for Nathan Vale" (documentation)
   → Note: Auth required, no content extracted

---
Created: 2 meeting notes, 1 clipping
Deleted: 3 log entries

Deleting Log Entries

After creating the note, delete the log entry from the daily note using Edit tool.

The raw transcription is preserved in the meeting note, so nothing is lost.

Creating Notes Efficiently

Step 1: Create note with content injection (single call)

para_create({
  template: "meeting",
  title: "GMS Project Kickoff - Jackie",
  dest: "00 Inbox",
  args: { ... },
  content: {
    "Attendees": "- Jackie (Team Lead)\n- Nathan",
    "Notes": "### Key Points\n- Point 1\n- Point 2",
    "Action Items": "- [ ] Task 1\n- [ ] Task 2"
  },
  response_format: "json"
})

Step 2: Set frontmatter (args may not populate correctly)

para_frontmatter_set({
  file: "00 Inbox/🗣️ Note Title.md",
  set: { meeting_type: "general", meeting_date: "2026-01-06", ... }
})

Step 3: Append Raw Transcription after Follow-up section

The Follow-up section has a - placeholder. Insert AFTER it:

para_insert({
  file: "...",
  heading: "Follow-up",
  content: "\n---\n\n## Raw Transcription\n\n> [transcription]",
  mode: "after"
})

Step 4: Update daily note - Use Edit tool directly (don't re-read via MCP).