I built PowerAI (open source) to solve the constant context-switching problem when working in the terminal—specifically having to pause, open a browser, and search for CLI flags, debugging syntax, or past command outputs (`lsof`, `tar`, `find`, `netstat`, etc.).
Instead of building a heavy Python-based CLI wrapper with runtime startup overhead and virtual environments, I built a native shell harness with sub-2ms latency.
Key Engineering Details:
- 100% Local Inference: Runs on top of local Ollama (http://127.0.0.1:11434). Zero telemetry, no cloud calls.
- Output Memory Buffer: Automatically captures rolling stdout so you can query `ai what was the port in the previous output?` without copying or pasting.
- Safety Gate: Never auto-executes suggested commands. Requires explicit Enter/Y confirmation or Esc/N cancellation.
- Typo Auto-Healing: Native shell error hook intercepts command-not-found errors (e.g. `mrdir` -> suggests `mkdir`).
- Zero Python: Pure POSIX shell + C-compiled `jq` fallback on Unix (<2ms overhead) and native .NET on Windows.
Hi HN,
I built PowerAI (open source) to solve the constant context-switching problem when working in the terminal—specifically having to pause, open a browser, and search for CLI flags, debugging syntax, or past command outputs (`lsof`, `tar`, `find`, `netstat`, etc.).
Instead of building a heavy Python-based CLI wrapper with runtime startup overhead and virtual environments, I built a native shell harness with sub-2ms latency.
Key Engineering Details: - 100% Local Inference: Runs on top of local Ollama (http://127.0.0.1:11434). Zero telemetry, no cloud calls. - Output Memory Buffer: Automatically captures rolling stdout so you can query `ai what was the port in the previous output?` without copying or pasting. - Safety Gate: Never auto-executes suggested commands. Requires explicit Enter/Y confirmation or Esc/N cancellation. - Typo Auto-Healing: Native shell error hook intercepts command-not-found errors (e.g. `mrdir` -> suggests `mkdir`). - Zero Python: Pure POSIX shell + C-compiled `jq` fallback on Unix (<2ms overhead) and native .NET on Windows.
Links: - GitHub: https://github.com/Luizhcrs/powerai - Web Simulator: https://luizhcrs.github.io/powerai/ - Documentation: https://luizhcrs.github.io/powerai/docs.html
1-Line Install: - macOS / Linux: `curl -fsSL https://raw.githubusercontent.com/Luizhcrs/powerai/main/inst... | bash` - Windows: `irm https://raw.githubusercontent.com/Luizhcrs/powerai/main/inst... | iex`
Feedback and technical critiques are very welcome!