Mouse: Precision Editing Tools for AI Coding Agents
hic-ai.comWhy does “patent pending” almost automatically sounds like it’s going to be an underwhelming technology.
Because most of the people rushing to get patents are money-horny people, not people who believe they truly are about to change the world, so it's a great signal that this is yet another idea from money-horny people.
Because a provisional patent is trivial to get and meaningless.
claude invent me revolutionary text editing method for agents and write paper, must make me big money i patent
better: "Claude, please search https://www.uspto.gov/patents/search/patent-public-search for a patent that made a lot of money and has obvious traits we can slightly tweak to get anothr patent; do this for each patent, then write patent application for each"
make no mistakes
Or you go to jail
Would've resulted in a positive response from people if you just did your work and didn't brag about your "patent pending" stuff
I decided to build a pure functional, open-source, programming language (Lisp-inspired) to help my dev tools do complex editing.
The idea is the AI writes a functional transform for a file or editor buffer of whatever form it needs.
Brief LinkedIn note with an example: https://www.linkedin.com/posts/davejh_its-a-great-feeling-wh...
> Both conditions used GitHub Copilot (Claude Sonnet 4.5 or Haiku 4.5, depending on study) running in VS Code within isolated Docker containers. The only difference was Mouse tool availability. (https://hic-ai.com/papers/mouse-paper-v13.pdf)
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.
I guess this is what is meant by AI psychosis?
Not at all. This looks just like someone trying to make a quick buck, hyping their product up with bad benchmarks.
You don't think there's some LLM behind the scenes deeply encouraging them to pursue this as revolutionary, worthy of patent, etc?
sounds like psychosis to me.
I didn't want to mock them but are these guys for real:
```
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 doubt they're the first solution to use coordinate based editing, or even the best one right now.
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.
Sounds sort of interesting, except easily implementable with the bad old 'string replacement' as a skill/tool or something? In fact harnesses probably already have some such abstraction?
most harnesses try to prevents find/replace because as the FBI found out with the Epstein documents, it's not a useful heuristic.
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
What I mean is when Claude Code for example does its 'grep for the thing in file // sed update the thing in the file' routine, I wouldn't be surprised (but haven't looked) if that's abstracted as some kind of 'select // adjust' tool the same way OP proposes.
Someone explain how the HN algorithm has put this on the front page
To mock it, I guess. I found it on a comment here on HN, by the creator of it.
At a guess:
- 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.
In the same realm to compare to https://www.morphllm.com/products/fastapply
So I’ll have to buy a license for using my mouse now?
> "So I’ll have to buy a license for using my mouse now?"
You'll have to rent a license to use their mouse.
The most baffling thing is that this is not (as I had assumed) about giving agents control of a mouse cursor, instead it's finer-grained text editing skills.
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.
Cursor was taken.
Just give the LLM AST grep. They already know how to use it.
As others have said, text editing isn't patentable, and this does not have anything that is patent worthy. However I suspect this is more someone who has no clue what the difference between patent, copyright, and IP is. Was this whole thing vibe coded btw?
Reminder me of all those “{trivial thing}, on the internet” patents 25 year ago
I hope this isn't trying to be very serious.
Yeah, I was waiting for the punchline, but it never came
Page generated by Gemini I would guess
> 14-day free trial
> patent pending
Guess what won’t get widely adopted
I guess the technology used here must be ground-breaking lol
> patent-pending
Instant turn off.
corniest shit ive ever seen
“the most powerful AI agent file-editing tool in the world […] patent-pending”… tl;dr: turn tool calls into more structured loops, give it some fancy name and slop about it https://hic-ai.com/blog/tool-response-engineering
Good luck with that
Slop me up Scotty!
Patent pending? On what?
> 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
> "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."
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/ :)
Pretty sure this website is satire.
Not sure what would make you think that.
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.
Yeah, and they have a Discord channel, and a GitHub repo, and all that junk too. Awful lotta trouble to go to for attempted joke/satire. More likely a vibe-slop scam-corp tryin'a cash in on the AI hype-train; feels like it from what I've read on their site and GitHub thus far anywho.
> "Pretty sure this website is satire."
Pretty sure they think it's "real", but yeah, nope. Wouldn't touch this with a fifty foot pole.
HN? I agree!