# iFlow CLI: A Terminal Coding Agent That Shuts Down on April 17, 2026

> iFlow CLI is a Node.js terminal agent that runs free hosted models from the iFlow open platform, with subagents, MCP installs and plan mode. The README states the service is winding down, so the decision is about migration, not adoption.

**iflow-ai/iflow-cli** — iFlow cli is a comprehensive command-line intelligence that embeds in your terminal, analyzes your repositories, does coding tasks, interprets your needs across contexts, and boosts efficiency by performing tasks from simple file operations to complex workflow automation.

- Repository: https://github.com/iflow-ai/iflow-cli
- Website: https://cli.iflow.cn/?
- Stars: 5,091 · Forks: 416
- Language: Shell
- License: not declared
- Published: 2026-09-23 · Updated: 2026-09-23 · Language: en
- Canonical page: https://hysenlabs.com/projects/iflow-ai-iflow-cli

## What iFlow CLI Actually Is, and Who It Was Built For

iFlow CLI is a command-line AI assistant that runs in your terminal, analyzes a repository, and performs coding tasks from file edits up to multi-step workflows. It is written primarily in Shell, and the repository is not archived, but its last push was on 2026-03-20, which is more than six months before today. The README opens with a notice that iFlow CLI will be shutting down on April 17, 2026 (UTC+8), and points to a farewell post for migration guidance. That single line reframes the whole review: this is not a project to build a five-year workflow on.

The intended audience is developers who want an agent in the terminal without paying per token. The README lists free access to models on the iFlow open platform, naming Kimi K2, Qwen3 Coder and DeepSeek v3. It also ships a feature comparison table against Claude Code and Gemini CLI, claiming Todo Planning, SubAgent, Plan Mode, Task Tools, a JetBrains plugin, conversation recovery, an open market, hooks, output styles and a workflow feature. Those are the project's own claims, not measured results, and the table is the kind of marketing artifact every CLI agent ships.

The practical question for a reader is narrower than the feature grid. If you already run an agent in your terminal, iFlow CLI's distinguishing pieces are the free hosted models and the built-in market for MCP servers and subagents. If you do not, the shutdown notice means the sensible move is to treat this as a short evaluation, not a migration target.

## How the Agent Runs: Modes, Context Compression and Subagents

The README describes four running modes that govern what the model may do: yolo, where the model has maximum permissions and can perform any operation; accepting edits, where it may only modify files; plan mode, where it plans first and executes after; and default, where the model has no permissions. That last one is the safety setting, and it is the mode to start in on an unfamiliar repository.

Context handling is explicit rather than hidden. The task tool is described as compressing context length so the CLI can finish longer tasks, with auto-compression triggered when context reaches 70%. Subagents turn the CLI into what the README calls an expert team, and the /agent command lists pre-configured ones. Conversation history can be saved and rolled back through iflow --resume and the /chat commands, which matters because the failure mode of a long agent session is usually a wrong turn taken twenty steps back.

The repository layout supports this picture: an IFLOW.md at the top level, plus docs_en/ and docs_cn/ directories holding the command reference. The README's own workflow is to run /init in an existing codebase, which scans the project and writes an IFLOW.md with documentation of its structure. That file is the agent's persistent memory of the repository, and it is also the artifact you would have to port if you moved to another tool.

## Installing iFlow CLI and Running Your First Task

The README requires Node.js 22 or newer, 4GB of RAM, and macOS 10.15+, Ubuntu 20.04+/Debian 10+, or Windows 10+ with WSL or Git for Windows. It states that Bash, Zsh or Fish work best. On macOS, Linux and Ubuntu the recommended path is a one-click script:

```bash
bash -c "$(curl -fsSL https://cloud.iflow.cn/iflow-cli/install.sh)"
```

Homebrew and npm are the alternatives. The npm package is @iflow-ai/iflow-cli, and the README notes the command installs the dependencies your terminal needs:

```bash
brew tap iflow-ai/iflow-cli
brew install iflow-cli
npm i -g @iflow-ai/iflow-cli
```

On Windows the README walks through downloading the Node.js installer from nodejs.org, running it, restarting CMD or PowerShell, then running npm install -g @iflow-ai/iflow-cli. For mainland China it documents an nvm route with mirror settings: nvm node_mirror https://npmmirror.com/mirrors/node/ and nvm npm_mirror https://npmmirror.com/mirrors/npm/, followed by nvm install 22 and nvm use 22 before the npm install.

Authentication comes next. The recommended option is iFlow's native login, which opens account authentication in a web page. The alternative is an OpenAI-compatible API connection, which the README suggests for servers where no browser is available. To get a key, you register an iFlow account, open profile settings or the direct link at iflow.cn/?open=setting, and click Reset in the dialog to generate one, then paste it at the terminal prompt.

With that done, launch the CLI from your workspace and let it read an existing repository:

```bash
cd project1/
iflow
> /init
```

The README states that /init scans the codebase, learns its structure, and creates an IFLOW.md with documentation. You should see that file appear at the top of the project. From there you can issue a task in plain language, for example asking it to analyze requirements from a requirement.md file, produce a technical document, and implement the solution. To remove the tool later, the README gives npm uninstall -g @iflow-ai/iflow-cli.

## The Shutdown Date Is the Real Constraint

The most important limitation is not technical. The README states plainly that iFlow CLI will be shutting down on April 17, 2026 (UTC+8). Any workflow you build on the hosted models, the open market, or the native authentication inherits that deadline. The repository itself is not archived, and the last push was on 2026-03-20, so there is no signal of active development after that date.

There are ordinary constraints too. Installation requires Node.js 22 or newer, which is a hard floor for machines pinned to older runtimes. Authentication and AI processing both need an internet connection, so the tool is not usable offline. The README does not document rollback of the auto-upgrade behaviour it mentions, nor does it describe what happens to your IFLOW.md, conversation history or installed MCP servers when the service ends. Those are the questions the farewell post is meant to answer, and the README defers to it.

Licence is another gap. The repository listing gives no licence, and the README does not state one. Without a licence file, the terms under which you may redistribute or modify the code are unclear, and that is a question for your own legal review rather than something this article can settle.

## iFlow CLI vs Claude Code and Gemini CLI

The README's own comparison table places iFlow CLI against Claude Code and Gemini CLI. The differences it claims are worth reading carefully because they describe approach rather than quality. Both Claude Code and Gemini CLI are marked as having limited multimodal capability in China, and Gemini CLI's search is marked as requiring a VPN, while iFlow CLI lists search and multimodal as available. Claude Code is listed as not free; Gemini CLI as limited usage; iFlow CLI as free.

The table also claims features the other two lack: a built-in open market, a workflow feature, and a JetBrains plugin. Claude Code is credited with Todo Planning, SubAgent, Plan Mode, Task Tools, hooks, output styles and thinking, so the overlap is substantial. The genuine difference in approach is distribution: iFlow CLI routes inference through the iFlow open platform with its own account system, while the alternatives route through their vendors' accounts. That is why the shutdown matters more here than it would for a locally configured agent.

Two other tools appear in search interest around this project: opencode and qwen code. The README does not describe either, so there is nothing here to compare them on beyond noting that readers are looking.

## Questions New Users Ask About iFlow CLI

The most common questions about this project are about installation, definitions, and how it stacks up against other terminal agents. The answers below come from the README and the repository layout, and they include the one fact that overrides the rest: the stated shutdown date of April 17, 2026 (UTC+8).

## Conclusion

iFlow CLI is worth installing only if you want a short, low-cost look at how a terminal agent handles plan mode, subagents and MCP before you commit elsewhere, or if you already depend on the iFlow open platform and need to inventory what has to move. Do not start a new long-lived workflow on it: the README states the shutdown date of April 17, 2026, and the repository's last push was on 2026-03-20. Before doing anything, open the farewell post at vibex.iflow.cn/t/topic/4819, check whether your authentication is the native iFlow login or an OpenAI-compatible key, and confirm what your IFLOW.md and MCP configuration files would need to move to another agent.

## FAQ

### How do I install iFlow CLI?

On macOS, Linux or Ubuntu the README recommends running bash -c "$(curl -fsSL https://cloud.iflow.cn/iflow-cli/install.sh)". Homebrew users can run brew tap iflow-ai/iflow-cli followed by brew install iflow-cli, and npm users can run npm i -g @iflow-ai/iflow-cli. Node.js 22 or newer is required.

### What is iFlow CLI?

It is a command-line AI assistant that runs in your terminal, analyzes code repositories and performs coding tasks, including file operations and multi-step workflows. It connects to free models on the iFlow open platform such as Kimi K2, Qwen3 Coder and DeepSeek v3, and supports subagents and MCP installs from the iFlow Open Market.

### How does iFlow CLI compare with opencode?

The README does not mention opencode, so there is no comparison in the project's own documentation. The README compares iFlow CLI against Claude Code and Gemini CLI on features such as plan mode, subagents, the built-in open market and free model access.

### How does iFlow CLI compare with qwen code?

The README does not mention qwen code, so no comparison is documented there. The only named comparison is the table against Claude Code and Gemini CLI, which claims iFlow CLI offers free access, a built-in open market and a workflow feature the others lack.

## Sources

- [iflow-ai/iflow-cli on GitHub](https://github.com/iflow-ai/iflow-cli)
- [Issues](https://github.com/iflow-ai/iflow-cli/issues)
- [Project website](https://cli.iflow.cn/?)
- [README](https://github.com/iflow-ai/iflow-cli/blob/main/README.md)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/iflow-ai-iflow-cli