Thank you for the feedback
@Pranshu @ Traycer Hey yes im currently using it and its been an adventure... I was looking for something that could tacke a 2000+ line code forcely created by github Copilot, and misguided by me (a new coder) and Traycer definately managed it, now..how would I call it. .. If Github Copilot is a Glock, somehting like Windsurf is a Ak47... Traycer is definately a .50Cal SNIPER.... You have to be evry careful, and very precise with what you want or it will go off the rails using random files that you ahvent even mentioned in cotext but theyre avalable at the root level... I like the review funciton but I dearly miss being able to talk to the reviewer (somethign like the githubcopilot chat) on something Im concerned about, not only what the Reviewer finds.. Honestly its been the only consistent app aside from copilot that has mademe think that I need to have a solid subcription on. Other things like it being capable of seeing your "problems" tab and fixing basic stuff would be nice because when i said the sniper reference, i meant that using it for very basic tasks is like using thor's hammer to kill a moskito. Aside from that very happy with the workflow up intul now, going to keep using it.
19 Replies
Thank you for sharing your experience and for taking the time to provide such detailed feedback! We’re happy 😊 to hear that Traycer has been able to handle your challenging codebase and that you find it powerful and consistent.
Out of curiosity 🤔, was the codebase relatively new when you started working on it? Since you mentioned being new to coding and dealing with files over 2000 lines long, Traycer likely attempted to find related files for context. It uses methods like analyzing file names, function references, definitions, regex searches, and more. Root-level files are typically considered only if they seem related. Let us know if this aligns with your experience!
Your suggestion for a chat feature is really interesting. Are you looking for something where you could ask the chat to review a specific portion of your code or clarify concerns about certain findings? We’d love to hear more about how you’d envision using this feature.
Thanks again for your thoughtful feedback—it’s incredibly valuable for helping us improve. If you have any more suggestions or ideas, feel free to share them. We’re excited to see how Traycer continues to support you in your coding journey!
@muziksociety,
Loved your analogies! We are big time 3D shooter fans as well! @LTDOGOOP is into mil-sims. @tesh and I are into CS and Pubg
Did you discover that every review comment allows replying back to AI for back and forth conversation?
Yes i did and i found it really interesting becaus e it the same type of conversation id like to have, if it has mycodebase in its grasp theres things a programmer might forget or might want t ask of its code specifically so an open coversation could work to refresh the mind of the programmer of how a function is tied to different files or how is the css implemented without having to actually go inide the code and explore it until finding the specific code. Im pretty sure youve seen the services that allow you to talk to a 30 page pdf.. for example, well that same principle.. if the code is already loaded and contexted inside of traycer, why not be able to talk to it, wether it may be for a consult, or a review on a specfic part that YOU choose, not traycer, or location specific feature suggestion.. a million things come to mind with an open conversation with your own codebase.
as a new coder one of my favorite things to do with cline or copilot even cursor when i tried it.. was to ask it to study my entire codebase BEFORE doing anything, because i have quetsions or suggestions and i need the coder tool to understand the entirety of my code for context aware questioning. Obviously, theres context window, tokens to consider here, but then again, theres a huge market of people who are coming in like me, people who have never coded but understand and are willing to duke it out with an AI to make MVP's or to fully have a coder that they can order arround, those people specially would find that chat with their code very useful, and it would be a nuke for people who actaully code because theyll be able to acces the code a lot faster and that would liberate mind space for other tasks.
And about the 2000 line file, i figures it was trying its best to find the information since my prompt was simple..
Got it. The Tasks is taking an interesting shape soon. At step 3 (code changes) the user can continue the conversation which will spin up a follow up plan. The follow up plan will have all the context from the previous tasks in the chain. This will turn it into a DMR (fast firing sniper) rather than a bolt action .50 cal
Yeah I saw that too, once you send the fix, it recalculates the plan right? But I cant see anywhere the input I applied after i send it, and if you do it again for something else, it forgets the first thing you asked because it goes abck to the original input + the last addition. Again, still this whole planning approach i had seen in in Plandex but the whole CLI thing was tough to handle and its definately mroe elegant and useful in traycer ( i know... ive tried a lot of the coding assistants)
once you send the fixWhat exactly do you mean by this? Step-3 changes are based on the plan that was finalized in Step-2. We don't really go back to it
I cant see anywhere the input I applied after i send it,We are adding quality of life improvements like apply-all/diff-all which will probably help with this?
Ok you send prompt 1:
1- creates a plan
2-displays Plan
3-you go to the bottom and talk to the code for any specific new fix, submit
4-The plan recalculates with your new suggestion
5-new plan displays, with added fixes
6-You go and add another suggestion at the bottom, submit
7- Plan forgets about #3 and continues to recalculate with #6
8- plan is displayed with #6 suggestion, and forgets about #3 suggestion.
i hope thats a lot clearer jaja
you go to the bottom and talk to the code for any specific new fixHow are you doing this right now? We are allowing just chat in the context of a single file
Ill show you
I guess you are talking about plan iteration prompt...
I see you are going back and forth from the generated changes to the plan
Yes

That right there
Got it. So, you are saying:
1. Original plan
2. Update with prompt1 -> Incorporated prompt1 changes
3. Update with prompt2 -> Incorporated prompt2 changes but misses prompt1 changes
yes exactly
This must be a bug we need to fix, cc: @Pranshu @ Traycer
BTW, you still need to hit generate changes in order to actually generate the code changes
Plan iteration prompts will just update the plan
Yeah
True
Well yeah this has been my experience until now, Again i come from a 0 coding background, but I have employed coders before and theyve always told me that even tho i dont know coding, I understand how to communicate with them which is usually very important for them apparently.. traycer ws a little overwhelming at first but everything is overwhelming for me at the moment since coding is a lot of information to take in at the same time. Up until now, when my 15 days are over you guys will definately have me around!
That is great to hear! We will love to have you as well, your feedback has been super helpful. Can't wait to hand the DMR in your hands!
You can think of Traycer as the Sniper's essential pack
Man, I excited! Once I finishe this browser extension im making, ill be posting in Reddit and ill definately plug Traycer as the gamechanger for me