Live data from Hacker News

Gemini CLI

blog.google

711–720 of 820 posts

Re: Gemini CLI

#711
post #41

Hi - I work on this. Uptake is a steep curve right now, spare a thought for the TPUs today. Appreciate all the takes so far, the team is reading this thread for feedback. Feel free to pile on with bugs or feature requests we'll all be reading.

Could Gemini CLI be used for PR reviews? For example, would you expect that asking Gemini to compare two git commit hashes and analyze the code changes for potential bugs/conciseness/adhesion to the idiomatic project style, etc. to work well?

Edit: I tried it. The setup was a breeze. I fed the CLI two git commit IDs and some light prompting on what to look for. It gave a reasonable response. I'll try on a real PR shortly.

Re: Gemini CLI

#712
post #670

Earlier quoted context omitted.

Not even "cargo install" is needed. Just `wget -O ~/.local/bin/gemini-cli https://ci.example.com/assets/latest/gemini-cli ` (Or the CURL version thereof) It can pick the file off github, some CI's assets, a package repo, a simple FTP server, an HTTP fileserver, over SSH, from a local cache, etc. It's so simple that one doesn't need a package manager. So there commonly is no package manager. Yet in this tread people a…

> is so simple, that you don't need a package manager! You might not know the reason ppl use package managers. Installing this "simple" way make it quite difficult to update and remove compared to using package managers. And although they are also "simple", it's quite a mess to manage packages manually in place of using such battle-tested systems

> You might not know the reason ppl use package managers.

People use package managers for the following:

- to manage dependencies - to update stuff to a specific version or the latest version - to downgrade stuff - to install stuff - to remove stuff

any of these, except for the dependency management, are a single command, or easy to do manually, with a single compiled binary. They are so simple that they can easily be built into the tool. Or handled by your OSs package manager. Or with a "shell script" that the vendor can provide (instead of, or next to, the precompiled binary.

I did not say manually, you infer that. But I never meant that. The contrary: because it's so simple, automating that, or have your distro, OS or package manager do this for you, is trivial. As opposed to that awful "curl example.com/install.sh | sudo tee -" or those horrible built-in updaters (that always start nagging when I open the app - the one moment that I don't want to be bothered by updates because I need the app now)

The only reason one would then need a package manager is to manage dependencies. But a precompiled binary like Go's or Rusts typically are statically compiled so they have no (or at most one) dependency.

Imagine the ease of a single ".targz" or so that includes the correct python version, all pips, all ENV vars, config files, and is executable. If you distribute that - what do you still need pip for? If you distribute that, how simple would turning it into a .deb, snap, dmg, flatpack, appimg, brew package, etc be? (Answer: a lot easier than doing this for the "directory of .py files. A LOT)

Re: Gemini CLI

#713
post #41

Hi - I work on this. Uptake is a steep curve right now, spare a thought for the TPUs today. Appreciate all the takes so far, the team is reading this thread for feedback. Feel free to pile on with bugs or feature requests we'll all be reading.

There is one feature in Claude Code which is often overlooked and I haven't seen it in any of the other agentic tools: There is a tool called "sub-agent", which creates a fresh context windows in which the model can independently work on a clearly defined sub-task. This effectively turns Claude Code from a single-agent model to a hierarchical multi-agent model (I am not sure if the hierarchy goes to depths >2). I won…

Would Gemini non-interactive mode be a stop gap if they don't have sub-agent equivalent yet?

https://github.com/google-gemini/gemini-cli/blob/main/docs/c...

Re: Gemini CLI

#714
post #41

Hi - I work on this. Uptake is a steep curve right now, spare a thought for the TPUs today. Appreciate all the takes so far, the team is reading this thread for feedback. Feel free to pile on with bugs or feature requests we'll all be reading.

Thanks for building this! The tool shows a lot of promise. Coming from Claude Code, the core functionality feels solid - just needs some permission refinements to match enterprise use cases. This is based upon quickly skimming the current code. High ROI feature requests: • Pattern-based permissions - Bash(git: ) to allow git but not rm, Write(logs/ .txt) for path scoping • CLI permission flags - --allowedTools "Read,…

On the one hand, yes this has obviously high immediate value; on the other hand, I can't help but feel like you are giving access to multiple tools that can be used for arbitrary code execution anyway (i.e. running tests, installing dependencies, or running any linter that has a plugin system...), so blacklisting `git --exec-path=/bin/sh` for example is... Misguided? You would have a better time containing the agent in an environment without internet access?

Re: Gemini CLI

#715
It worked yesterday, but stopped today with "Failed to login. Ensure your Google account is not a Workspace account. Message: Resource has been exhausted (e.g. check quota). "

I guess I will use something else. This is all very annoying given that I actually pay for Gemini Pro...

Re: Gemini CLI

#716

Earlier quoted context omitted.

There is one feature in Claude Code which is often overlooked and I haven't seen it in any of the other agentic tools: There is a tool called "sub-agent", which creates a fresh context windows in which the model can independently work on a clearly defined sub-task. This effectively turns Claude Code from a single-agent model to a hierarchical multi-agent model (I am not sure if the hierarchy goes to depths >2). I won…

Would Gemini non-interactive mode be a stop gap if they don't have sub-agent equivalent yet? https://github.com/google-gemini/gemini-cli/blob/main/docs/c...

Possibly. One could think about hooking this in as a tool or simple shell command. But then there is no management when multiple tools modify the codebase simultaneously.

But it is still worth a try and may be possible with some prompting and duct tape.

Re: Gemini CLI

#717

Earlier quoted context omitted.

Thanks for building this! The tool shows a lot of promise. Coming from Claude Code, the core functionality feels solid - just needs some permission refinements to match enterprise use cases. This is based upon quickly skimming the current code. High ROI feature requests: • Pattern-based permissions - Bash(git: ) to allow git but not rm, Write(logs/ .txt) for path scoping • CLI permission flags - --allowedTools "Read,…

On the one hand, yes this has obviously high immediate value; on the other hand, I can't help but feel like you are giving access to multiple tools that can be used for arbitrary code execution anyway (i.e. running tests, installing dependencies, or running any linter that has a plugin system...), so blacklisting `git --exec-path=/bin/sh` for example is... Misguided? You would have a better time containing the agent…

It’s not misguided. The goal isn’t prefect security, the goal is mitigating risk and collaborating with cross functional security, compliance, platform, operations, etc… teams.

Use Jules, also by Google if you need what you describe.

Re: Gemini CLI

#718
In google colab, how are we supposed to run it?

Well, not sure that it makes sense to do it, anyway I've tried to run in in a cell and in the google colab terminal. Still waiting for auth (?)

Re: Gemini CLI

#719
I just get downgraded to 2.5 Flash after like three prompts and can't seem to find an option to switch back. So much for "60 requests per minute are free".

Re: Gemini CLI

#720
post #400

Can't we standardize on AGENTS.md instead of all these specific CLAUDE.md and now GEMINI.md. I just symlink now to AGENTS.md

I agree that we should standardize on AGENTS.md but Gemini CLI does at least let you configure the filename: https://github.com/google-gemini/gemini-cli/blob/3a369ddec3b...

Claude Code seems to require the CLAUDE.md filename.

https://technicalwriting.dev/ai/agents/

Post reply on HN