Explore spaceStart a project

Neel CLI

Your AI workflow. In your terminal.

Your AI workflow in the terminal, with the models you choose and permissions you control.

The Neel CLI is the terminal flagship: a coding agent that reads, searches and edits your project under the permissions you give it, with local models first and a receipt you can open for any turn.

Join the waitlist Email us
Early accessIn development · terminal workflow
~/projects/atlas · zshIllustrative
neel inspectrun      turn 7 · Complete · 2.1 smodel    Selected local model · on your deviceturn     Complete · 3,507 in · 82 out · 39.8 tokens/s · 2 model calls · 1 tool call · 2.1 s call 1   Request context · tool requested · 1,704 in · 18 out · 44.0 tokens/s · 0.4 s└ read_document launch-brief.md · succeeded · 12 ms · allowed oncecall 2   Compose response · completed · 1,803 in · 64 out · 39.8 tokens/s · 1.7 s sent     system prompt · 1 tool definition · request bodyreceived raw response · 64 tok · finish stopkept     saved with the session · survives a restart · exportable
Illustrative session. Values show the receipt format, not a benchmark.

Inside the terminal

Your tools. Your permissions. Your pace.

The features below exist in the current development build, with early access still ahead.

  • A receipt for every turn

    neel inspect shows the turn line, every model call and every tool call, and exports the full receipt.

  • Approvals that fit

    Allow or deny a tool call once, or for every matching call, and a matching choice is kept for eight hours.

  • Plan and build

    Switch modes with Shift+Tab. Plan mode withholds every tool that could change your project.

  • Your project rules

    Reads your AGENTS.md, so the agent follows the conventions you already wrote down.

  • Headless when you need it

    Run a single prompt with -p and get JSON or JSONL back for scripts and CI.

  • Honest model picker

    See how much memory each model needs and how much is free before it loads.

  • MCP servers

    Add, list and manage MCP servers from the terminal.

  • Endpoints for other tools

    An ACP agent and an OpenAI-compatible local endpoint, in developer preview.

  • Shell completions

    Completions for Bash, Zsh and Fish.

Models and hardware

Local first, any runtime.

Run MLX, GGUF through llama.cpp or LiteRT-LM locally, point the CLI at a local server you already run, or add a hosted provider you opt into.

  • Curated defaults

    Gemma

    A guided starting point

    Neel’s curated Gemma 4 options come with documented memory guidance for the initial local runtime.

    Gemma 4 E2B · E4B · 12B
  • Bring your own model

    Qwen

    Coding and multilingual work

    A broad family for code, language and reasoning tasks, with smaller and larger variants to explore through a compatible local runtime.

    Qwen3 · 4B · 8B · 14BAbout Qwen (opens in a new tab)
  • Bring your own model

    Llama

    Compact everyday assistants

    Smaller text models for conversation, summarising and rewriting. A useful family to explore when you want a lighter local setup.

    Llama 3.2 · 1B · 3BAbout Llama (opens in a new tab)
  • Bring your own model

    DeepSeek

    Reasoning-focused experiments

    The R1 distilled models offer smaller options for reasoning tasks. Pick the exact variant and context size for the memory you have available.

    DeepSeek-R1 Distill · Qwen and Llama variantsAbout DeepSeek (opens in a new tab)

These are model families to explore, not a list of Neel-tested builds. Beyond the curated defaults, bring a compatible GGUF model or connect a supported local server. Compatibility, tool use and memory needs depend on the exact model file, runtime and context length; not every variant is supported.

Curated Gemma memory guide

These documented defaults cover the Gemma 4 MLX 4-bit configuration on macOS with Apple silicon. This picker covers these three models only; it is not a cross-platform hardware guarantee.

Example device memory

Start with Gemma 4 E4B. A comfortable starting point with 16 GB in this configuration.

Curated Gemma 4 MLX models and the free memory each needs
ModelRoleFree memory neededDevice memory
Gemma 4 E2B3.5 GiB on diskDefault6.5 GiB8 GB (tight)
Gemma 4 E4BComfortable8.5 GiB16 GB
Gemma 4 12BLargest local22.5 GiB32 GB

These figures apply to the curated Gemma 4 MLX 4-bit variants on macOS, not every model family or runtime. Neel checks available memory before loading. Apple's on-device model offers another starting point where it is available, without a separate model download.

Privacy defaults

Quiet, local and yours.

  • No account for local use

    Local models run without signing in and without an inference bill.

  • Models download only when you choose

    Neel ships without model weights. Nothing large downloads without your say.

  • Every download SHA-256 verified

    Runtimes fetched on first use are pinned and checked against a SHA-256.

  • Uninstall keeps your data

    Uninstalling removes only what we installed. Models, settings and conversations stay yours.

Status and install

How it will install.

The CLI launches with Neel Desktop, the editor extensions and Neel Mobile, planned for January 2027, on Windows, macOS and Linux, with no Rust, Swift or Python toolchain needed.

  • Homebrew on macOS and Linux, with Windows install steps published at launch.
  • An offline installer that checks a downloaded archive against SHA256SUMS before it installs anything, keeps versions side by side and can roll back.
  • Never a pipe-to-shell script. Every release lists its SHA-256, third-party notices and release notes.

Join the waitlist and we will send setup details when early access is ready.

Status
Early access · In development · terminal workflow
Launch platforms
Windows, macOS, Linux
Launch
January 2027, with the whole ecosystem
Account
Not needed for local models
Binary
neel

Early access

Get the Neel CLI first.

Join the waitlist for early access, or email us about what you want to build with Neel. We will get in touch when it is ready for your machine.

Join the waitlist Email us