Synapse-Shell is a lightweight bridge that connects your Linux terminal directly to local AI. It treats an AI model as a powerful command-line utility—similar to 'grep' or 'awk'.
It takes the output from any command (via pipes) and sends it to a local LLM with your instructions. The AI processes that data and returns the result directly to your terminal.
- Privacy: 100% local via Ollama. No data leaks.
- VRAM Efficient: Purges GPU memory immediately after the task.
- Unix-Native: Built for pipes and automation.
- Explain Load Spikes:
(ps aux --sort=-%cpu | head -10) | s 'Who is eating my CPU and should I kill it?' - Storage Cleanup:
df -h | s 'Identify the most critical partitions and suggest cleanup commands' - System Health:
uptime | s 'Is this load average normal for a 4-core CPU?' - Network Audit:
ss -tulpn | s 'Which of these open ports are potential security risks?'
- Brute-Force Detection:
tail -n 50 /var/log/auth.log | s 'Find failed login attempts and group them by IP' - Kernel Debugging:
dmesg | tail -n 20 | s 'Explain these kernel errors in simple terms' - Docker Audits:
docker logs my_container --tail 20 | s 'What is causing this container to crash?'
- Refactor Code:
cat script.py | s 'Rewrite this using list comprehensions for better performance' - Generate Docs:
cat function.go | s 'Write a professional Javadoc-style comment for this' - Unit Testing:
cat api.py | s 'Write 3 edge-case tests for this endpoint using pytest' - SQL Optimization:
echo "SELECT * FROM users WHERE id > 10" | s 'Is there a more efficient way to write this?'
- Format Conversion:
cat data.csv | s 'Convert this into a clean Markdown table' - Summarization:
cat long_notes.txt | s 'Give me the 5 most important bullet points' - Translation:
cat readme_de.md | s 'Translate this into professional English'
- Explaining Flags:
echo "tar -xzvf archive.tar.gz" | s 'What does each flag (x, z, v, f) actually do?' - Command Discovery:
ls --help | s 'How do I sort this output by file size?'
- Have Ollama running.
- Install:
./install.sh - Use the shortcut:
cat log.txt | s 'Analyze this'
Author: Axel Jerabek