Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

DeepWiki: Understand Any Codebase

aitidbits.ai

Post reply on HN
Print view View original

31–40 of 55 posts

Re: DeepWiki: Understand Any Codebase

#31

Post by tacker2000 »

Nullabillity wrote: post #24
tacker2000 wrote: post #5

The Elkjs project uses this and im not really sure i like it. [1] Its a bit hard to find stuff. I was looking to find the structure of the main configuration json object and couldnt find it in the deepwiki. I found it on the “non ai created” doc page of the main Elk project[2] (Elkjs is a JS implementation of Elk) But yes this is of course just one data point. [1] https://deepwiki.com/kieler/elkjs/5-usage-guide [2] h…

"Uses it" sounds strong.. I don't see any link to it from https://github.com/kieler/elkjs ? Annoyingly, anyone can just.. request a deepwiki for any GitHub repo. That one exists doesn't mean that it's endorsed or reviewed by the project. They just kind of barged in, welcome or not. Just another SEO slop-spammer.

Hmm you are right, there is no direct link.

I seem to have found it via a google search.

So in the end people will believe that these are the official docs…

Re: DeepWiki: Understand Any Codebase

#32

Post by buovjaga »

IshKebab wrote: post #27
johnfn wrote: post #26

Earlier quoted context omitted.

Out of curiosity, how come LibreOffice has .buckversion, BUCK, .buckconfig, etc? This commit[1] does seem to indicate using Buck to build at one point, though it is 10 years old. [1]: https://github.com/LibreOffice/core/commit/1fd41f43eb73c373c...

Indeed and the BUCK file was updated 2 months ago. It definitely does use BUCK. Maybe it's not the main build system, but it's not hard to see why someone or something unfamiliar with the project would think that. Really there should be a comment in the BUCK file explaining what it is.

Ok, I submitted a patch with a comment.

Re: DeepWiki: Understand Any Codebase

#34

Post by debugnik »

Nullabillity wrote: post #24
tacker2000 wrote: post #5

The Elkjs project uses this and im not really sure i like it. [1] Its a bit hard to find stuff. I was looking to find the structure of the main configuration json object and couldnt find it in the deepwiki. I found it on the “non ai created” doc page of the main Elk project[2] (Elkjs is a JS implementation of Elk) But yes this is of course just one data point. [1] https://deepwiki.com/kieler/elkjs/5-usage-guide [2] h…

"Uses it" sounds strong.. I don't see any link to it from https://github.com/kieler/elkjs ? Annoyingly, anyone can just.. request a deepwiki for any GitHub repo. That one exists doesn't mean that it's endorsed or reviewed by the project. They just kind of barged in, welcome or not. Just another SEO slop-spammer.

A couple of years ago the OCaml and Julia languages already had to deal with a content farm that created wikis for them, filled them with LLM-generated, blatantly wrong or stupidly low quality content, and SEOed its way above actual learning materials. Cue in the newbies to these languages being incredibly confused.

This at least tries to generate the text out of the actual project, but I'm pessimistic and think it'll cause similar confusion.

Re: DeepWiki: Understand Any Codebase

#35

Post by caboteria »

I recently received an AI-slop bug report for a small open source project (PureLB) that I maintain, and the slop was generated by DeepWiki. It was very incorrect, but I didn't know what "DeepWiki" was so I wasted about an hour. If DeepWiki is causing garbage bug reports even on tiny projects like mine, I can't imagine how much maintainer time it's wasting over all.

Re: DeepWiki: Understand Any Codebase

#36

Post by letaem77 »

lelouch9099 wrote: post #33

What if I don't trust this third party with my code? are there any open source/ local way to run this ?

This is my way to do it:

1. Archive whole repository into single text file with Repopack: https://github.com/yamadashy/repomix

2. To reduce token, compress the file with LLMLingua-2: https://github.com/microsoft/LLMLingua

(fewer token = more context can be given to LLM = LLM better understands your repository)

3. Copy & Paste the compressed archive text contents as a context, into ChatGPT’s input field as-is, or local LLMs.

4. Ask the LLM for documentation generation. Like, “this is a repository source code. given context, generate a ‘table-of-content’.” Then you will get a ToC. If it looks good, you can ask for generating first chapter. And keeps going until you finish whole documentation.

If you are trying to document Typescript/Javascript codebase, You may use bundlers like esbuild for step 2, which will beneficial for token reducing.

If you interested in step 2’s LLMLingua-2, check out my Typescript port that can be ran without no installation at: https://atjsh.github.io/llmlingua-2-js/

Re: DeepWiki: Understand Any Codebase

#37

Post by baby »

buovjaga wrote: post #23

It's a pity that there is no clear way to send takedown requests. We didn't ask for deceptive garbage to be generated as documentation for LibreOffice, but here it is and newbies are discovering it: https://deepwiki.com/LibreOffice/core/2-build-system (spoiler: LibreOffice has never used Buck as a build system)

From my experience using deepwiki what it generates is not deceptive garbage

Re: DeepWiki: Understand Any Codebase

#38

Post by giancarlostoro »

rgoulter wrote: post #4

The DeepWiki tool itself seems pretty neat. It has a pretty good go at collecting documentation from across the codebase and organising it in one place. It has a pretty good guess at coming up with documentation which isn't there. It strikes me as an example of automated code assistance that's e.g. more useful than "the item under the cursor has type , here's its documentation". There are things which benefit from be…

I'm trying it out with a repo that has no code comments or documentation. Let's see how it does. :)

Edit: its been like over 10 minutes, and still nothing, thats interesting, I did choose a Lingo source project so it probably gave up by now.

Re: DeepWiki: Understand Any Codebase

#39

Post by conartist6 »

buovjaga wrote: post #23

It's a pity that there is no clear way to send takedown requests. We didn't ask for deceptive garbage to be generated as documentation for LibreOffice, but here it is and newbies are discovering it: https://deepwiki.com/LibreOffice/core/2-build-system (spoiler: LibreOffice has never used Buck as a build system)

I sent them a politely worded threat and they responded right away opting me out:

> Hello, I am writing you as an author of Open Source software seeking to protect my security and that of my users.

> What I would like to know is: how may I prevent deepwiki from indexing my projects, specifically those in the ----- GitHub organization? If you consider yourselves to have implicit legal permission to train on my projects and write about them, know that I hereby explicitly and permanently revoke that permission.

> Since you likely believe that I lack the authority to get you to stop, I will add this:

> To the extent allowed by law I will consider any incorrect information you publish about my projects to be libelous and, given this notice, made with your intention. LLMs have no will to act, so publishing misinformation about my project, at such time as that happens, could only be the result of human will.

> Kind regards, > Conrad Buck

Re: DeepWiki: Understand Any Codebase

#40

Post by michaelmior »

conartist6 wrote: post #39
buovjaga wrote: post #23

It's a pity that there is no clear way to send takedown requests. We didn't ask for deceptive garbage to be generated as documentation for LibreOffice, but here it is and newbies are discovering it: https://deepwiki.com/LibreOffice/core/2-build-system (spoiler: LibreOffice has never used Buck as a build system)

I sent them a politely worded threat and they responded right away opting me out: > Hello, I am writing you as an author of Open Source software seeking to protect my security and that of my users. > What I would like to know is: how may I prevent deepwiki from indexing my projects, specifically those in the ----- GitHub organization? If you consider yourselves to have implicit legal permission to train on my project…

Just because you "consider" incorrect information to be libelous does not mean that it is. While it's true that LLMs have no will to act, the use of an LLM to publish information that ends up being incorrect does not imply that the user of the LLM intended to post incorrect information.
Post reply on HN