home / skills / krishagel / geoffrey / obsidian-manager
This skill helps you manage your Obsidian vault, read and write notes, and search highlights for rapid knowledge retrieval.
npx playbooks add skill krishagel/geoffrey --skill obsidian-managerReview the files below or copy the command above to add this skill to your agents.
---
name: obsidian-manager
description: Manage Obsidian vault for persistent knowledge storage, search, and retrieval
triggers:
- "save to obsidian"
- "search my notes"
- "find in vault"
- "check my readwise"
- "search readwise"
- "search snipd"
- "check my highlights"
- "create note"
- "add to obsidian"
- "what did I read about"
- "what podcasts mentioned"
allowed-tools: Read, Write, Bash, Glob, Grep
version: 0.1.0
---
# Obsidian Manager Skill
Manage the user's Obsidian vault as a persistent second brain. Read existing knowledge, write new content, and search across years of highlights and notes.
## Vault Configuration
**Path:** `~/Library/Mobile Documents/iCloud~md~obsidian/Documents/Personal_Notes/`
**Write Mode:** Auto-create new files, confirm before updating existing files.
## Vault Structure
```
Personal_Notes/
├── Geoffrey/ # Geoffrey-generated content
│ ├── Mission Control.md # Central dashboard (Dataview queries)
│ ├── Scheduled Tasks.md # Scheduler dashboard
│ ├── Research/ # Research task outputs
│ │ ├── Daily AI Briefings/ # Daily automated research
│ │ ├── {Project}/ # Other research projects
│ │ └── attachments/ # Research images
│ ├── Reports/ # Generated reports and artifacts
│ │ ├── PSD/ # Peninsula School District reports
│ │ │ ├── Discipline/ # Discipline reports and infographics
│ │ │ ├── Enrollment/ # Enrollment reports
│ │ │ ├── Tech Support/ # Tech support dashboards
│ │ │ └── {topic}/ # Other PSD topics
│ │ ├── HRG/ # Hat Rack Group (consulting LLC) reports
│ │ │ └── {topic}/ # HRG topics
│ │ ├── Personal/ # Personal reports
│ │ │ └── {topic}/ # Personal topics
│ │ └── {org}/ # Other organizations as needed
│ ├── Daily-Logs/ # Session summaries (YYYY-MM-DD.md)
│ ├── Learnings/ # Extracted patterns (YYYY-MM-DD-topic.md)
│ └── Decisions/ # Major decisions with rationale (YYYY-MM-DD-topic.md)
├── Meetings/ # Meeting notes (YYYY-MM-DD format)
├── People/ # Contact profiles with metadata
├── Readwise/ # Synced highlights (368 articles, 34 books)
│ ├── Articles/
│ ├── Books/
│ ├── Podcasts/
│ └── Tweets/
├── Snipd/ # Podcast highlights (58+ episodes)
│ └── Data/ # Podcast folders > episodes
└── Templates/ # Note templates
```
**CRITICAL:** See CLAUDE.md "Obsidian Organizational Rules" section for:
- Strict folder structure requirements
- File naming conventions
- Routing rules by content type
- Validation checklist before writing
Always validate folder placement before writing any files to Geoffrey folder.
## Scripts
All scripts use Python via `uv run` with inline dependencies.
### Search Scripts
#### search.py - General Vault Search
```bash
uv run scripts/search.py "query" [folder] [--limit N]
```
Search across the vault for keyword matches. Returns file paths and matching excerpts.
**Examples:**
```bash
# Search entire vault
uv run scripts/search.py "prompt engineering"
# Search specific folder
uv run scripts/search.py "leadership" People
# Limit results
uv run scripts/search.py "AI" Readwise/Articles --limit 10
```
#### search-readwise.py - Search Highlights
```bash
uv run scripts/search-readwise.py "query" [--author "name"] [--category articles|books|podcasts]
```
Search Readwise highlights by topic, author, or category.
**Examples:**
```bash
# Search by topic
uv run scripts/search-readwise.py "second brain"
# Search specific author
uv run scripts/search-readwise.py "AI" --author "Andrej Karpathy"
# Search books only
uv run scripts/search-readwise.py "habits" --category books
```
#### search-snipd.py - Search Podcast Transcripts
```bash
uv run scripts/search-snipd.py "query" [--show "podcast name"]
```
Search Snipd podcast transcripts and snips.
**Examples:**
```bash
# Search all podcasts
uv run scripts/search-snipd.py "machine learning"
# Search specific show
uv run scripts/search-snipd.py "leadership" --show "Huberman Lab"
```
### Creation Scripts
#### create-note.py - Create New Note
```bash
uv run scripts/create-note.py "title" "content" [--folder Geoffrey/Research] [--tags tag1,tag2] [--related "[[Note]]"]
```
Create a new note with proper frontmatter and optional backlinks.
**Examples:**
```bash
# Create research note
uv run scripts/create-note.py "AI Model Comparison" "Content here..." --folder Geoffrey/Research --tags research,ai
# Create with backlinks
uv run scripts/create-note.py "Meeting Notes" "Content..." --folder Meetings --related "[[John Smith]],[[Project Alpha]]"
```
#### open-in-obsidian.py - Open Note in Obsidian
```bash
uv run scripts/open-in-obsidian.py "path/to/note.md"
```
Opens the specified note in Obsidian using Actions URI.
## Content Routing
**CRITICAL:** When creating content, use these destinations. NEVER dump files directly into `Geoffrey/` - always use proper subfolders.
| Content Type | Folder | Naming Pattern |
|--------------|--------|----------------|
| Research results | `Geoffrey/Research/` | `YYYY-MM-DD-topic.md` |
| **Reports** | `Geoffrey/Reports/{org}/{topic}/` | Descriptive name with date |
| **Images/Infographics** | Same as parent report | `Descriptive Name YYYY-MM-DD.png` |
| Meeting notes | `Meetings/` | `Topic - YYYY-MM-DD.md` |
| Daily summaries | `Geoffrey/Daily-Logs/` | `YYYY-MM-DD.md` |
| Learnings | `Geoffrey/Learnings/` | `topic-slug.md` |
| Decisions | `Geoffrey/Decisions/` | `YYYY-MM-DD-decision.md` |
| Person updates | `People/` | Existing file name |
**Naming Rules:**
- All images/infographics MUST include date in filename
- Format: `Descriptive Name YYYY-MM-DD.png` or `Descriptive Name YYYY-MM-DD to YYYY-MM-DD.png` for date ranges
- Example: `Peninsula High School Discipline Report 2025-11-17 to 2025-11-20.png`
**Reports Folder Structure:**
Pattern: `Geoffrey/Reports/{organization}/{topic}/`
Organizations:
- **PSD** - Peninsula School District reports
- `Geoffrey/Reports/PSD/Discipline/`
- `Geoffrey/Reports/PSD/Enrollment/`
- `Geoffrey/Reports/PSD/Attendance/`
- `Geoffrey/Reports/PSD/{topic}/`
- **HRG** - Hat Rack Group (consulting LLC) reports
- `Geoffrey/Reports/HRG/{topic}/`
- **Personal** - Personal reports
- `Geoffrey/Reports/Personal/{topic}/`
- **{org}** - Other organizations as needed
## Frontmatter Standard
All Geoffrey-created notes should include:
```yaml
---
created: 2025-11-23
tags: [geoffrey, research]
source: geoffrey
related:
- "[[Related Note 1]]"
- "[[Related Note 2]]"
---
```
## When to Read from Vault
- **Before research tasks:** Check if user has relevant Readwise highlights or existing notes
- **When people mentioned:** Look up context in People folder
- **For meeting prep:** Find past meetings and related notes
- **For context:** Search Snipd transcripts for podcast discussions on topic
## When to Write to Vault
- **After research:** Save results to `Geoffrey/Research/`
- **Learned patterns:** Extract insights to `Geoffrey/Learnings/`
- **Major decisions:** Document with rationale in `Geoffrey/Decisions/`
- **Session summaries:** Optional daily logs to `Geoffrey/Daily-Logs/`
## Creating Backlinks
Always create backlinks to existing content:
1. **Search before creating:** Find related notes first
2. **Use wiki-links:** `[[Note Title]]` format
3. **Add to frontmatter:** List in `related:` field
4. **Reference sources:** Link to Readwise/Snipd sources when relevant
**Example:**
```markdown
Based on insights from [[Readwise/Articles/Building a Second Brain]],
and discussed in [[Snipd/Data/Huberman Lab/episode-123]],
the key pattern is...
```
## Readwise Content Format
Readwise notes have this structure:
```yaml
---
Author: "Author Name"
Full Title: "Article Title"
Category: #articles
Summary: "AI-generated summary"
---
## Highlights
- Highlight text with optional tags like [[topic]]
```
## Snipd Content Format
Snipd episodes have rich metadata:
```yaml
---
episode_show: "Podcast Name"
episode_publish_date: 2024-01-15
mentioned_books: ["Book 1", "Book 2"]
snips_count: 5
---
### Snip 1
**Timestamp:** 00:15:30
> Quote text here
**Transcript:** Full context...
```
## Actions URI Integration
Use Actions URI to interact with Obsidian app:
```bash
# Open a note
open "obsidian://actions-uri/note/open?vault=Personal_Notes&file=Geoffrey/Research/note.md"
# Create and open
open "obsidian://actions-uri/note/create?vault=Personal_Notes&file=path&content=..."
```
## Best Practices
1. **Search before creating:** Always check for existing content
2. **Link generously:** Create connections between notes
3. **Use consistent tags:** Match existing tag patterns
4. **Keep notes atomic:** One idea per note in Learnings
5. **Include sources:** Always cite Readwise/Snipd sources
6. **Respect existing structure:** Don't reorganize user's folders
## Output Format
When completing Obsidian operations:
```markdown
## Summary
What was done
## Files
- Created: `path/to/new.md`
- Updated: `path/to/existing.md` (if confirmed)
## Related Content Found
- [[Relevant Note 1]] - why it's relevant
- [[Relevant Note 2]] - why it's relevant
## Next Steps
- Recommendations if any
```
This skill manages an Obsidian vault as a persistent second brain for reading, writing, searching, and linking notes. It automates safe file creation and supports disciplined routing, naming, and frontmatter standards so content remains discoverable and consistent. Use it to surface context before tasks and to persist research, decisions, and daily summaries.
The skill reads from the vault to find relevant notes, Readwise highlights, and Snipd transcripts using targeted search scripts. It creates new notes with validated folder placement, standardized frontmatter, and wiki-style backlinks, and it can open notes in Obsidian via Actions URI. All writes validate routing rules and confirm updates to existing files.
How do writes avoid breaking the vault structure?
Writes validate folder placement and naming rules before creating files; they auto-create appropriate subfolders and confirm before updating existing files.
How are backlinks handled?
The skill searches for related notes first, inserts wiki-links in content, and lists related notes in the frontmatter related field to preserve connections and discoverability.