realrocker

I am working on speeding up the review-refine loop of LLM generated artefacts between me and the LLM. i run my coding agent (claude code) in auto / full-permission mode, so it doesn’t stop for accept/reject — by the time it’s done i’ve got a pile of generated code, docs, sometimes ui to review. and the slow part for me is to refine it before push it to remote. i would copy-paste snippets to point at what i wanted changed, screenshot the ui, describe where the problem was in prose, one message at a time. its cumbersome.

so i built prereview to speed up that review-fix loop. you run it in your repo (or point it at a file, a dir, or a running dev server), click what’s wrong: a diff line, a markdown/html block, a region of an image, a box on a live site — and leave a comment. the comments go to a csv, your agent reads it and fixes things and tells you what it did via a comment or even posts suggestions which you can accept or reject. it ships with a claude code skill, but the handoff to llm is just an open csv protocol, so any llm cli can drive it. stuff that might interest people here: single static go binary, fully offline; it reviews docs/images/live ui as well as code; comments re-anchor when the file changes under them. it’s mit and still early but I use it everyday. Its here: https://github.com/livetemplate/prereview

I am using claude cli/tmux on a hetzner box and connecting to it via claude remote control. I have connected the box and my phone over telnet which allows me to view any UI work. Sometimes I do have to switch to my laptop for UI desktop layouts.

One gap which I kept running into on both mobile and desktop was refining the initial plan and then later refining the generated artifacts which involved lots of imprecise copy-paste. To scratch my own itch I built a review tool to improve the velocity of planning and refining generated artifacts. It has become my daily driver: https://github.com/livetemplate/prereview

exactly. LLM code output is so high that it has become unreasonable to review the code in terminal or ide

Yeah the project is right out of the oven so will add some usage gifs shortly to the repo. Meanwhile here are some screenshots on imgur: https://imgur.com/a/zxZo651

:) We are all converging on the same problems and tooling to fix them.

Built this in the last couple days to review changes done by claude on a hetzner box(over tailscale on my phone) and leave comments before pushing to remote : https://github.com/livetemplate/prereview.

Runs everywhere and can has a claude skill: /prereview which starts a new instance for the current changes. Has a “Hand-off” button to signal claude to process the comments.

UI is responsive. I use it on my iphone with claude mobile app + /remote-control to leave comments before pushing.Still requires polish but core features work.

that’s exactly the core value proposition of the tool I have been working on : https://github.com/livetemplate/tinkerdown

I am working on https://github.com/livetemplate/tinkerdown to leverage the the two effective outputs of LLMs : html and markdown . Still WIP

Recently started using: https://github.com/babonet/Markco (cant find the similar extension i used previously since it got uninstalled). I add inline comments in the markdown and I have a skill which uses the comments to update the plans and remove resolve comments.

Good ideas. I am working on something similar but with a tiered complexity to fallback to html for complex use cases : https://github.com/livetemplate/tinkerdown, https://github.com/livetemplate/tinkerdown/blob/main/docs/gu...

excellent work. I have been using a vscode extension for this for an year.

No that makes sense. Waiting for feedback might lead to churn. Pretty cool idea.

That is super cool. Sorry to be nitpicky but would really like to know your mental model: I didn’t understand from the blog why user waiting for a functional UI is a problem ? isn’t the partial streamed UI non-functional ?

I can see the value in early user verification and maybe interrupting the LLM to not proceed on an invalid path but I guess this is customer facing so not as valuable.

"In interactive assistants, that latency makes or breaks the experience." Why ? Because user might just jump off ?

(edited)

The streamed execution idea is novel to me. Not sure what’s it significance ?

I have been working on something with a similar goal:

https://github.com/livetemplate/tinkerdown

Last time I wrote here was almost 6 years ago: https://honestmusings.wordpress.com. Not sure when I will write again.

The first step counter button is supposed to reload the page. The second step buttons update the page w/o reload

Hey. I have checked out pushup before. Its great. I guess I am taking a niche approach which not many would find useful. To clarify: What is adding more complexity? alpinejs ?

Its locking and the state is shared. Didn’t expect a lot of interest for this or I would have built a better demo :)

htmx is great!

The (claimed) value proposition is leveraging block/template to decompose the html page and bind its computation to browser events over a pubsub channel. Its a toolkit and not a framework so one can still use their favourite framework. The Go library isn’t doing anything earth shattering but re-implementing the aforesaid design would still need to be done in any framework.

The state is shared between users for the demo. Its possible to separate state in the library per user but I haven't enabled it.

Yeah I was afraid of that. Websockets are costly. Thankfully the approach doesn't completely depend on websockets so I have disabled it for now on the site.

It is intended as a longer-term project originating from the previous experiment: https://github.com/adnaan/gomodest. You are right, a better selection criteria for the target audience should be enlisted. I have been working on this for a while so I just got greedy for some early feedback.

This example has most of the scenarios you have called out: https://github.com/livefir/fir/tree/main/examples/fira. Fir aims to limit itself to rendering templates on the server and making it available for all subscribers as a browser CustomEvent which is consumed by alpine.js for more complex interactivity. The expectation is the that the developer handles it via either alpine.js plugins or standard JS code.

[edit] This is still a work-in-progress so I will certainly add better guides and documentation moving forward. Although its good to get feedback that this approach might be interesting for at least some people.

The app was on a 256mb instance. I have scaled it to 2gb. Let's hope its sufficient for now

An in-situ todo list in every tool I use. In slack: scheduled messages, email: todo folder, git: todo.txt and so on. I no longer fight context switching and focus on getting things done in an organised way every time I switch context to a new task.

To write complex code while I am procrastinating or context switching I have taken to writing in plain english what I want to achieve and then write code. It’s cheaper to get all the bad ideas out in plain english while making reasonable progress.

Been working on a html over the wire library in Go: https://github.com/livefir/fir. It mostly works but still work in progress

My prompt: A planet where story is currency used for buying food

https://twitter.com/adnaanx/status/1598757303922065431?s=46&...

Like you said in another thread, there are two choices here. I took the other choice because I wanted to: 1. not build a new templating language 2. not maintain a custom server side renderer(v8). 3. Keep javascript to the minimum 4. Build for developers who know Go and maybe a little html/css. 5. No special html attributes. It should be just standard html/template i.e easy to delete the websocket stuff if it doesn’t work for you 6. Offer a view library instead of a framework. Go users are likely to be more open to adopting a library.

The approach is similar to hotwire except I only use stimulusjs and not use turbojs. turbojs doesn’t integrate well with html/template. In my approach, you can simply reuse the html templates you have already written with standard html/template package.

Sorry for the rambling, I haven’t gotten around to writing down all of it in a more structured way.

Oh wow someone finally went ahead with the server rendered js in Go ! Looks great. I have been experimenting various ways to build web apps in Go: https://github.com/adnaan/gomodest-template. I have landed on pursuing one approach more deeply: server rendered html templates over websockets. Don’t have a lot of documentation right now but here are some examples: https://github.com/goliveview/examples.

Agreed. I have been playing with the idea of escaping to use svelte in spots for finer interactions : https://github.com/adnaan/gomodest-template

The more businesses move to SaaS models, automation takes a different shape as well, i.e automation scripts involving business objects within/outside the system.

Well put. I see this getting applied to cloud infrastructure too. Cloud providers are increasingly offering their products as pure software objects on which traditional software engineering principles can be applied. I have been experimenting with using typescript to define infra components for last 18 months and have seen higher code contributions from non-operators(most of them are typescript practitioners) w.r.t terrafrom or ansible.

[slightly shameless plug] Recently started using Deno + typescript and its even better. My thoughts on this here: https://adnaan.badr.in/blog/2021/11/20/deno-for-infrastructu...