Haiku/Sonnet 4.5 on GitHub Copilot is not a valid comparison whatsoever.
You need to benchmark against Claude Code running Opus. I mean, being revolutionary is a big claim to fame.
```
Instead of:
"Update the checklist to mark items 1.2, 1.4, and 1.5 as done."
"Mark items 1.2, 1.4, and 1.5 as complete in the checklist. Only insert an x in each checkbox. Do not copy or replace any of the item descriptions."There is not universe in which this would make agents more efficient - and who is prompting their agents like that in the first place?
I also asked glm to extract all the tools and tell me how they work roughly and nothing interesting really just slop:
```
The server exposes exactly 11 tools (verified via the xa whitelist at L16918, not the larger Eo metadata map which contains ~24 tool definitions — most are dead/legacy):
- 6 read/meta: read_first_n_lines, read_last_n_lines, read_lines, jump_to_line_n, find_in_file, get_file_metadata
- 3 edit/control: quick_edit (6 ops: insert/delete/replace/replace_range/for_lines/adjust), batch_quick_edit (atomic, always-staged, max 500 ops, multi-file), save_changes, cancel_changes
- 1 always-on: license_status
Notable design choices:
- Coordinate-based addressing (line/char/rect) instead of content-echo — saves tokens
- Staging model: edits go to an in-memory shadow, save_changes is the only disk mutation
- The rect + move "click-and-drag" columnar editor (v0.9.7) is the genuinely novel bit
- ReDoS static analyser (~700 lines) protects find_in_file ```
I believe openode uses a tool that requires line numbers and the model needs to write the entire line again. It also prevents it from making changes if it hasn't read the document since it was last changed.
There's certainly improvements, but no, the simple cli commands arn't properly made to do what a dumb model wants to do, since the dumb model isn't ever going to consider there's a bunch of other Donalds or Trumps or DTs that are completely innucuos
Eg: Check out hash-anchored editing. The first place where I recall seeing this was the oh-my-pi coding agent, but I wouldn't be surprised if the idea originated earlier/elsewhere.
I wonder whether CRDTs could be a good solution for multiple agents editing the same codebase in parallel.
- No "bad history" from submitter.
- No detected "obvious slop" signs
- Relatively (near zero) few comments during first hour, during which time it received steady, unclustered, unique upvotes.
- No actual mod took a look and weighted it either way
HN algo weights against rapid fire comment trees (sign of "controversy / chat" rather than thoughtful content (sort of)), obvious bot activity, upvotes from sketchy sources, etc - other than that submissions are pretty much bound for front page if they get a rate of organic votes.
You'll have to rent a license to use their mouse.
The word mouse has had an established meaning in computing for over half a century, so it seems like an odd term to lay claim to for something so unrelated.
Good luck with that
> insert a line, delete a range, replace a character, edit a column
The ed(1) command set 50 years old. I doubt it's patentable. These guys are far from the first to apply fine-grained text editing to LLM toolsets. I've been teaching models to do it for years. Hell, models want to use sed and awk so much that you have to hold them back.
I'm so repulsed by the idea that these guys think they can fence off a slice of the ancient commons, claim they discovered it, and charge $15/month to access it that I want nothing to do with them and will go to the mattresses to make sure they can't. Nobody owns text editing, not even when it's an AI doing it.
Mouse: sincerely, fuck you
Yeah, I been givin' Qwen a "toolchain" containing `sed`,`awk`,`rg`, and `git` in a "sandbox" directory for playin' around with text editing lately. Havin' a ton of fun dinkin' around with Ollama, Python, and Qwen. Don't need much more'n that to get yerself into all kinda trouble. ;)
Qwen and Gemma make a real fine pair with a bit of Python "glue" too. Gemma's real good with image data (classifying and describing, tagging, title-ing, extracting and translating text, etc) and Qwen's better at code related stuff, so... Teamwork, yay! \o/ :)
https://hic-ai.com/papers/mouse-paper-v13.pdf seems like an awfully lot of trouble to go through for a joke that isn't even funny.
HIC AI is a Delaware corporation, registration number 10476082, incorporated 1/16/2026.
Pretty sure they think it's "real", but yeah, nope. Wouldn't touch this with a fifty foot pole.