I've been frustrated with AI command-line tools that just spit out a single command and call it done. So I built Ask-Shell - an AI agent that actually completes your entire task.
Links
- GitHub: https://github.com/fssqawj/ask-shell
- PyPI:
pip install askshell-ai
What makes it different?
Other tools:
- Generate ONE command → You run it → Done (even if task isn't complete)
- Command fails? → You figure it out
Ask-Shell:
- Executes MULTIPLE steps automatically
- Analyzes results after each step
- Learns from failures and retries
- Doesn't stop until task is actually complete
Example
Task: "Organize my project files"
Shell-GPT: ls -la (one command, you do the rest)
Ask-Shell:
- Analyzes directory structure
- Creates organized folders
- Moves files to appropriate locations
- Verifies everything is in place ✓ Task complete!
Features
- 🔄 Multi-step task automation with intelligent loop
- 🧠 Automatic error recovery and retry
- 💭 Real-time AI thinking display
- 🔒 Dual-layer safety (AI + blacklist protection)
- 🎯 Interactive/auto/demo modes
Top comments (0)