local-terminal-llm¶
ask is a local-first terminal workflow built around the llm CLI, optional local retrieval, and shell-native command execution.
This site documents how to install it, configure it, and use its higher-level features safely.
What You Get¶
- A Python CLI (
askllm) normally invoked asask - Normal Q&A mode with selectable models
- Command-only mode (
-co) with optional direct execution (-no) - Retrieval grounding:
--webfor current web context--docsfor local document collections--autofor heuristic routing- Text-to-speech with Piper (
--tts) infullorstreammode - Shell integration for bash, zsh, fish, and similar shells
- Configurable prompt-insert hotkeys for
ask,codex,claude, or custom commands
Quick Start¶
./scripts/setup.sh
ask "Explain rsync -avz"
Only install/validate dependencies:
./scripts/setup.sh --dep-only
Typical Flows¶
# command generation only
ask -co "find the 20 largest files"
# command generation + execute after confirmation
ask -co -no "rename all .jpeg files to .jpg in this repo"
# stream speech while tokens are generated
ask --tts --tts-configure stream "summarize this changelog"