Inside the Core Engine
Want to back the stream? The «داعم» membership on YouTube is EGP 60 a month: badges, emoji, and I answer your questions first on stream. Become a supporter
The gist
The stream centers on developing the Burnish Shopify app, especially refining its multi-step onboarding so it feels native and consistent with Shopify, communicates merchant value clearly, and improves copy, cards, navigation, and pricing using visual references. The work remained unfinished: content and design revisions, backend wiring, deep-link behavior, and onboarding’s dependency on database state still required attention. In parallel, Luqman explores a safer Git workflow for multiple coding agents—cleaning the repository, isolating tasks with branches and worktrees, detecting conflicts before merge, validating development servers and URLs, and maintaining a living diagnostic log of difficult problems and fixes. Audience questions and broader discussions cover SEO and GEO, AI-driven product discovery, using funnel signals to assess demand, and practical advice for learning while building a SaaS product. The stream also contains personal, qualified impressions of Codex, Claude, Gemini, GPT, and coding benchmarks, stressing that outcomes depend partly on how tools are used and that informal comparisons do not support firm conclusions.
109 moments from the stream
- 1:06:51
GitHub Access Does Not Automatically Fix the Branching Workflow
The speaker discusses how the lack of a clear workflow slowed work across multiple coding sessions and tools. He wants to inspect old or untracked work and clean the repository cautiously after a merge problem rather than delete unclear files. He also explains that GitHub integration inside Codex may simplify viewing and creating branches, but access alone does not solve mistakes in coordinating branches, pull requests, merges, and conflicts. He remains uncertain about permission depth and some connection options while testing them.
- 1:21:52
Checking a Pull Request Before Merge and Detecting Conflicts
After accepting the deep-link visual QA despite incomplete optimization, Luqman discusses the need for a system that checks changes before merge or push and allows or rejects them. He explains that GitHub exposes code conflicts in pull requests, notes the Git/GitHub learning curve, and then encounters a conflict that should have been spotted earlier.
- 1:25:51
AI Performance Depends on How It Is Used
Luqman describes AI as a mirror of its user: someone more skilled at using it may move much faster, so outcomes depend not only on the model but also on the operator. He also relays an unverified claim he has heard that an unclearly named tool may be good at security; no test is shown.
- 1:27:58
Separating Unrelated Work Across Branches
Luqman explains that mixing landing-page and application changes in one branch creates confusion and makes the codebase harder to work with. He proposes finishing the current piece, cleaning up the state, and restarting with dedicated branches and stronger workflow rules.
- 1:30:46
Confirming Frontend Scope Before Merge
Before merging and pushing, Luqman asks for explicit confirmation about whether the pull request contains any design or frontend changes, helping define the review scope.
- 1:35:18
Learning a Git Workflow Through Practice
The speaker explains that he deliberately accepted extra complexity to practise branches, worktrees, and pull requests before needing that workflow later. One lesson he drew was to clean up the repository state before starting substantial work, then use dedicated branches and worktrees for separate tasks.
- 1:40:29
How AI Platforms Are Changing Product Discovery
The speaker notes that some shoppers have begun asking AI tools where to buy products alongside traditional search, connecting this shift with GEO and shopping features offered by some AI platforms.
- 1:42:29
Separating the Scan Funnel from the Shopify App
The discussion clarifies that the scan funnel and PDF report belong to the website and landing page rather than the Shopify app, so they should be retained while unnecessary experimental app routes and elements are removed.
- 1:43:54
Advice for a Graduate Who Wants to Build a SaaS Product
Answering a telecommunications engineering graduate, the speaker recommends using AI as a partner to explore ideas and next steps, then learning through practical building instead of waiting to complete a full study path. This is personal advice rather than a detailed roadmap.
- 1:45:22
Workflow Complexity and Harder Future Features
The speaker observes that most current problems concern the application workflow rather than the core app, while expecting some later advanced features to be difficult to implement. One feature name is unclear in the captions.
- 1:49:37
A Demo Mode for Multiple Application States
A demo system is needed to inspect the app and website from different perspectives because stores can vary in catalog size, visibility state, and performance score; a single data state would not reveal the full experience.
- 1:50:58
Using Customer Acquisition Cost to Evaluate a Funnel
The speaker is interested in measuring the funnel's customer acquisition cost and believes early advertising results may influence later product-direction decisions. No private figures or financial details are included.
- 1:53:36
A Clean Git Branch and Worktree Workflow
A discussion of starting from an up-to-date main branch, completing a milestone in an isolated workspace, verifying it, and then merging while minimizing clashes with other branches and worktrees. This is presented as a desired workflow, not one proven successful on this page.
- 2:00:03
Per-Worktree Development Servers and Falling Back to Main
An explanation that a development server can be launched from main or from a specific worktree to inspect that worktree’s changes before merging. When unexplained errors persisted, the proposed practical reset was to stop the servers and launch one from main; the page neither identifies the cause nor confirms a fix.
- 2:02:48
Cleaning Up Stale Git Branches Without Losing Work
A practical review of old and apparently unmerged branches: first verify whether the work is already on main, then push missing work or remove branches confirmed as merged, while avoiding abandonment of prior work that was never integrated.
- 2:11:37
Delegating Cleanup Through a Context-Preserving Conversation Fork
The speaker describes forking the current conversation so another agent retains the context of the problems already examined and can finish the cleanup while the main discussion moves on to preventive workflow design.
- 2:14:21
A Simple, Reliable Branch and Worktree System
A request for a global workflow that prevents untracked work, unknown or stale branches, and overlapping worktrees while several product areas are developed in parallel. The desired system should be simple, clear, easy to push and merge, usable by both Codex and Claude, and supported by a practical or theoretical explanation; earlier instructions in a shared file were acknowledged as insufficient.
- 2:20:11
Do Internal Skills Need Branches and Worktrees?
The speaker questions whether an internal coding tool or skill truly needs its own branch, worktree, and merge to main, or whether it belongs directly in a local repository for internal use. The issue is presented as a case the proposed workflow should resolve, with no final decision shown on this page.
- 2:23:13
A Practical Workflow for Parallel Development Sessions with Git Worktrees
A practical example considers one session working on application onboarding while another handles settings. The proposed workflow checks branch, file, and worktree state before creating an isolated worktree for each task, aiming to prevent changes from colliding. The speaker considers a modest increase in token usage worthwhile if it reduces time and confusion.
- 2:28:40
Can One Shopify App Dev Server Serve Multiple Worktrees?
The discussion raises the inconvenience of running multiple Shopify App Dev environments to inspect work from different worktrees. It asks whether a single server on the main branch could show the results of other sessions, but these captions do not document a clear final answer.
Got a question about something in this stream?
Ask in the group