claude-skills

Files (64)

• marketplace.json · .claude-plugin/
• plugin.json · .claude-plugin/
• .gitignore
• CHANGELOG.md
• CONTRIBUTING.md
• INSTALLATION.md
• INTEGRATION_COMPLETE.md
• INTEGRATION_TESTS.md
• LICENSE
• PACKAGE_SUMMARY.md
• PAPER_REVIEW_SYSTEM_CHANGELOG.md
• PROJECT_SUMMARY.md
• README.md
• RELEASE_NOTES_v1.0.0.md
• SKILL.md
• YOUTUBE_TRANSCRIBER_COMMANDS.md
• ky-orchestrator.md · agents/
• quant-accuracy-calibrator.md · agents/trading/
• quant-backtester.md · agents/trading/
• quant-risk-manager.md · agents/trading/
• quant-signal-generator.md · agents/trading/
• claude_code_commands.md
• archive-transcripts.md · commands/
• daily-note.md · commands/
• extract-transcript-facts.md · commands/
• format-transcript-notes.md · commands/
• integrate-obsidian-vault.md · commands/
• release.md · commands/
• run-youtube-transcriber.md · commands/
• scan-mr.md · commands/
• scan-pb.md · commands/
• todo.md · commands/
• update-memory.md · commands/
• validate-trade.md · commands/
• vault-check.md · commands/
• CHANGELOG_TEMPLATE.md · docs/
• README_TEMPLATE.md · docs/
• SKILL_TEMPLATE.md · docs/
• VERSION_GUIDE.md · docs/
• integration_test.py
• CHANGELOG.md · skills/memory/world-weaver/
• README.md · skills/memory/world-weaver/
• ww-bio-auditor.md · skills/memory/world-weaver/agents/
• ww-cache-analyzer.md · skills/memory/world-weaver/agents/
• ww-hinton-validator.md · skills/memory/world-weaver/agents/
• ww-leak-hunter.md · skills/memory/world-weaver/agents/
• ww-memory.md · skills/memory/world-weaver/agents/
• ww-orchestrator.md · skills/memory/world-weaver/agents/
• ww-race-hunter.md · skills/memory/world-weaver/agents/
• ww-retriever.md · skills/memory/world-weaver/agents/
• ww-synthesizer.md · skills/memory/world-weaver/agents/
• ww-trace-debugger.md · skills/memory/world-weaver/agents/
• audit.md · skills/memory/world-weaver/commands/
• consolidate.md · skills/memory/world-weaver/commands/
• context.md · skills/memory/world-weaver/commands/
• diagnose.md · skills/memory/world-weaver/commands/
• recall.md · skills/memory/world-weaver/commands/
• remember.md · skills/memory/world-weaver/commands/
• stats.md · skills/memory/world-weaver/commands/
• visualize.md · skills/memory/world-weaver/commands/
• …and 4 more

SKILL.md (first lines)

# YouTube Transcriber Pipeline - Claude Code Skill

**Version**: 1.0.0 | **Status**: Ready for Claude Code Integration

## Overview

Complete 4-skill pipeline for extracting, formatting, organizing, and archiving YouTube transcripts. Integrates with Claude Code CLI using the system's configured API keys (no manual key management needed).

## Capabilities

This skill provides a complete workflow:
1. **Extract Facts** - AI-powered fact extraction from transcripts (uses Claude API via Claude Code)
2. **Format Notes** - Convert facts to atomic markdown notes with wiki-links
3. **Integrate Obsidian** - Import notes into Obsidian vault with backlinks
4. **Archive Transcripts** - Safe archival with integrity verification

## Usage in Claude Code

```bash
# Run complete pipeline
/run-youtube-transcriber --transcript ~/Downloads/transcript.txt

# Just extract facts
/extract-transcript-facts ~/Downloads/transcript.txt

# Just format notes
/format-transcript-notes ~/Downloads/notes/ --output ./formatted

# Just integrate Obsidian
/integrate-obsidian-vault ./notes/ --vault-path ~/Documents/Obsidian/Aaron

# Just archive
/archive-transcripts ~/Downloads/transcript.txt
```

## Features

### Phase 1: Extract Facts (AI-Powered)
- Uses Claude 3.5 Sonnet via Claude Code CLI
- 7 fact categories (scientific, quotes, methods, examples, statistics, warnings, connections)
- Hierarchical topic organization
- Confidence scoring (0.0-1.0)
- Automatic metadata (actionable, philosophical, controversial)
- **No API key needed** - Uses Claude Code's configured authentication

### Phase 2: Format as Atomic Notes
- 1 fact = 1 markdown file
- Wiki-link cross-references [[Topic Name]]
- YAML front matter with metadata
- Topic indices and hierarchies
- Obsidian vault compatibility

### Phase 3: Integrate Obsidian
- Safe vault backup before changes
- Preserve directory structure
- Auto-generate backlinks
- Master index creation
- Link verification
- Dry-run mode for testing

### Phase 4: Archive Transcripts
- tar.gz compression
- MD5 integrity verification
- Metadata tracking
- Optional cleanup
- Verify-only mode

## Installation

The skill is ready to use once copied to Claude Code:

```bash
# Copy to Claude Code skills directory
cp -r ~/github/astoreyai/claude-skills/ ~/.claude/skills/youtube-transcriber

# Or individual components
cp -r youtube-transcript-extractor ~/.claude/skills/
cp -r transcript-to-logseq ~/.claude/skills/
cp -r transcript-to-obsidian ~/.claude/skills/
cp -r transcript-archiver ~/.claude/skills/
```

## Requirements

### For Claude Code CLI
- Claude Code installed and configured
- API key already configured in Claude Code (no manual setup needed)
- Python 3.11+

### For Obsidian Integration
- Obsidian vault at `~/Documents/Obsidian/Aaron/` (configurable)
- Read/write permissions to vault directory

## Architecture

```
Claude Code CLI
    ↓
Transcript Extractor (Phase 1)
    ├─ Uses Claude Code's API key automatically
    ├─ Extracts facts via Claude API
    └─ Outputs Fact objects
    ↓
Logseq Formatter (Phase 2)
    ├─ Converts facts to atomic notes
    └─ Creates markdown files
    ↓
Obsidian Integrator (Phase 3)
    ├─ Imports notes to vault
    ├─ Generates backlinks
    └─ Creates indices
    ↓
Transcript Archiver (Phase 4)
    ├─ Compresses files
    ├─ Verifies integrity
    └─ Stores archive
```

## Future: Ollama Integration

From astoreyai/claude-skills, MIT.

More from astoreyai/claude-skills