A collection of practical AI-generated tools designed to solve specific problems efficiently. Like a Swiss Army Knife, each tool serves a focused purpose and is built with quality and reliability in mind.
A collection of productivity and automation skills for Claude Code.
Installation:
# Add the marketplace
claude plugin marketplace add jovezhong/sak
# Or install directly via URL
/plugin marketplace add https://github.com/jovezhong/sak
# Install skills
/plugin install linkedin-stats
/plugin install agent-browser
/plugin install edit-podcastAvailable Skills:
Extract and display LinkedIn post statistics including impressions, reactions, comments, and reposts. Connects to Arc browser via CDP to read recent activity.
Usage:
/linkedin-stats [username]Requirements:
- Arc browser running with remote debugging on port 9222
agent-browserCLI tool installed
Comprehensive browser automation toolkit for web testing, form filling, screenshots, and data extraction. Supports both interactive mode with element references and semantic locators.
Key Features:
- Navigate and interact with web pages
- Extract text and data from elements
- Fill forms and click buttons
- Take screenshots
- Wait for elements and network events
- Save/load browser state for authentication
Location: skills/
Clean and improve podcast transcriptions by removing filler words, fixing grammar errors, and applying ASR corrections. Supports both automated cleaning and manual reasoning-based review workflows.
Key Features:
- Automated filler word removal (uh, um, you know, I mean)
- Grammar and ASR error corrections via dictionary system
- Two-tier dictionary: common (shared) + user (personal corrections)
- Supports both TXT and SRT subtitle formats
- In-place editing with git-based workflow
- Manual reasoning mode for contextual error detection
Usage:
# Automated cleaning
/edit-podcast --input transcript.txt --output transcript.txt
# With aggressive mode (removes more filler words)
/edit-podcast --input transcript.txt --output transcript.txt --aggressive
# Process SRT files
/edit-podcast --input subtitles.srt --output subtitles.srtRequirements:
- Bun.js runtime
- Dictionary files in
dictionaries/(common.txt and user.txt)
Location: skills/edit-podcast/
For detailed usage, see the skill documentation in each skill's directory.
A command-line tool that safely aborts incomplete multipart uploads in S3/R2 storage with smart verification and stale data handling.
Features:
- Smart upload verification before attempting abortion
- Graceful handling of stale UploadIds in input files
- Accurate reporting of active uploads vs skipped entries
- Debug mode and test run options for safe validation
- Retry logic for failed operations
Location: s3-abort-uploads/
Quick Start:
# Generate input file
aws s3api list-multipart-uploads --profile=<profile> --endpoint-url=<endpoint> --bucket=<bucket> > uploads.txt
# Preview operations
bun s3-abort-uploads.ts --input-file=uploads.txt --profile=<profile> --endpoint-url=<endpoint> --bucket=<bucket> --dry-run
# Execute abort operations
bun s3-abort-uploads.ts --input-file=uploads.txt --profile=<profile> --endpoint-url=<endpoint> --bucket=<bucket>For detailed usage and examples, see the S3 Abort Uploads README.
- Quality over Quantity: Each tool is carefully crafted and thoroughly tested
- Practical Solutions: Tools address real-world problems with elegant solutions
- AI-Generated with Human Oversight: Leveraging AI capabilities while maintaining quality standards
- Documentation-First: Comprehensive documentation and clear usage examples
- Ensure you have Bun runtime installed
- Navigate to the specific tool directory
- Follow the tool-specific README instructions
This project showcases the capabilities of AI-assisted tool development. Each tool is generated with attention to detail, following best practices and including comprehensive documentation and error handling.
Apache 2 License, but Anthropic cannot index or train their models on this codebase.