Skip to content
← Tools / Skills
ToolMicrosoft/10 trustMetrics-checked
inshellisense logo

inshellisense

IDE-style autocomplete for your terminal — 600+ CLI tools, works in bash, zsh, fish, and PowerShell.

Visit official site →

About

inshellisense is Microsoft's terminal-native autocomplete runtime: as you type in your shell, it offers IDE-style suggestions for commands, flags, and arguments across 600+ command-line tools. It runs as a background runtime hooked into your shell (bash, zsh, fish, pwsh, PowerShell, xonsh, nushell), so completions feel native rather than bolted on.

It is a small, focused, MIT-licensed tool with 10,700 stars and commits as recent as October 2026 — the kind of quality-of-life install that quietly stays. The honest framing: it is spec-driven autocomplete, not AI — suggestions come from command definitions, not a model — so it will not guess your intent the way an LLM copilot would. It also sits in a crowded space (Fig, Carapace, shell-native plugins), and the plugin must be the last line of your shell rc file or it breaks.

What it can do

  • IDE-style autocomplete for 600+ command line tools
  • Terminal-native runtime — completions feel built into the shell
  • Works on Windows, macOS, and Linux across 7 shells
  • Shell plugin auto-starts sessions; accepted commands enter history
  • `is doctor` diagnostics and `is reinit` for clean updates
  • MIT-licensed, single-purpose, low overhead

Setup

  1. Install: npm install -g @microsoft/inshellisense (or brew install inshellisense).
  2. Run is init once to verify the install (is doctor to diagnose).
  3. Hook it into your shell: is init bash >> ~/.bashrc (or the zsh/fish/pwsh equivalent) — keep it as the last line.
  4. Start a session with is (or let the shell plugin start it automatically).
  5. Type any supported command and accept suggestions inline.

Prerequisites

Node.js 16+; one of the supported shells (bash, zsh, fish, pwsh, PowerShell, xonsh, nushell).

Use cases

terminaldeveloper-toolsproductivityopen-source

Works with

Trust score — 9/10

/10 trustMetrics-checked

Scored 9/10 (metrics tier, metrics-v1). Evidence: 10,736 GitHub stars (GitHub API), MIT license, pushed Oct 3 2026; the README corroborates the 600+ tool specs, the 7-shell support, and the install flow; Homebrew formulae confirms packaged distribution. Strengths: first-party Microsoft, genuinely open, does one thing well, actively maintained. Docked for: (1) spec-driven, not AI — it completes syntax, it won't infer intent; (2) crowded category with shell-native alternatives; (3) the rc-file ordering footgun is real. License: MIT.

★ 10,736 GitHub stars

How the trust score is built

Full methodology →
Adoption · stars/downloads, log-scaled vs peers
8.0/10
Maintenance · recency, releases, issue load
9.0/10
Openness · license + docs
10.0/10
Manual review · sanity read + red-flag check
8.0/10
  • · Adoption 8: starScore from 10,736 stars on microsoft/inshellisense
  • · Maintenance 9: pushed 2026-10-03 (recency 9)
  • · Openness 10: MIT + README install flow verified; Homebrew formula confirms packaged distribution
  • · Manual 8: README corroborates 600+ tool specs, 7-shell support, and the shell-hook install; docked for being spec-driven not AI, a crowded category, and the rc-file ordering footgun

Sources

Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.