AutoShell: commands, data pipelines, and AutoScript
AutoShell, invoked as ash, brings everyday commands, structured pipelines, and AutoScript into one execution engine. Keep an interactive session open, or use the CLI to execute a command or script file.
It connects exploration with repeatable work: inspect files or data, filter and transform fields, then turn useful operations into a script.
All applications · Usage guide and recorded examples
Keep working in one session
Directories, variables, and aliases persist between commands. Completion, history search, and inline suggestions reduce repeated typing. Pipelines, redirection, command chains, and background jobs help compose daily work.
For example, browse a directory with ls, filter relevant files, and export results to another program. Interactive use and ash -c share the engine, making explored commands reusable in automation.
Pass structured records between commands
Some built-in commands return records with fields rather than only a text stream. File records from ls and parsed JSON can be filtered, sorted, projected, and converted.
cat users.json | from_json | .age > 30 | select .name .ageThis reads JSON, filters by age, and retains names and ages. The pipeline examples provide data, copyable commands, and recorded output. External-command output and structured records are not always the same data kind; handle them according to the command.
Extend a command into a script
F2 opens the multiline AutoScript editor for functions, conditions, and loops. Enter inserts a line; F5 executes and returns to the prompt. Ctrl+Enter depends on the terminal reporting its modifiers.
Capture query results in a script, then calculate, iterate, or summarize. Save a .ash file to repeat the operation through the CLI. The guide contains complete classification, amount-summary, and JSON-query examples.
Commands, scripts, and the AI entry
| Entry | Purpose |
|---|---|
| F1 | Lock shell command mode |
| F2 | Edit and execute multiline AutoScript |
| F3 | Enter AI chat input; a working model service configuration is required |
F3 opens an AI session. An input-area screenshot does not prove an external model call completed. AutoMusk organizes more complex development tasks; AutoAI explains shared model services.
Automation and execution scope
CLI options provide JSON output, read-only operation, execution previews, path and command restrictions, and audit records. Their coverage differs: policy applies to file operations managed by ash, while external programs require OS isolation for their own file access.
Current agent integration can use command calls, JSON results, and execution policy. Some dedicated Agent subcommands in older documentation are not implemented and are not current entry points.
Getting started and current boundaries
Building requires Rust and matching repository dependencies. Add the built ash to PATH, then use:
ash
ash --json -c 'ls'
ash user-report.ashRecorded runs cover the Windows CLI and interactive shell on 2026-09-29/30, using ash v0.1.0. Other platforms and standalone GUI/TUI coverage need separate confirmation. AutoLang's website release v0.5 and the ash executable version are identified separately.
The usage guide contains build instructions, current query syntax, script downloads, and observed limitations such as CSV output. This introduction was organized on 2026-10-01.
Real terminal and operation examples · AutoOS · Application overview