Working on Merchant Facing UI
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 follows preparations for a Shopify app submission, prioritizing completion of the move from Polaris React to Polaris Web Components and visual and functional QA of the merchant interface before returning to website, report, and funnel polish. Practical work also covered GA4, Google Tag Manager, and PostHog configuration; verifying that changes reached production through GitHub and Cloudflare; reviewing Supabase table organization; and troubleshooting MCP access—without treating visible identifiers alone as proof of end-to-end tracking. Luqman drew lessons from accumulated Git branches and worktrees and from coding-agent completion claims that failed manual inspection, arguing for regular integration and direct verification. He also discussed automated build-and-audit agent loops, his personal impressions of Codex and Claude, and audience questions about deployment, model choice, project languages, client acquisition, programmatic SEO, Kanban boards, and Hermes. Open questions remained around some Cloudflare and MCP failures, the best database organization, the value of particular agent tools, and the final submission date.
104 moments from the stream
- 0:47:39
An Automated Build, Review, and Dependency-Unblocking Cycle
The speaker explains a workflow for a dependency-linked milestone: a builder agent implements a prepared task, an independent auditor reviews it, and approval marks it complete and automatically unblocks dependent tasks. The cycle repeats until the milestone is finished, while the speaker remains skeptical of the agent’s time estimate.
- 0:50:39
Organizing a Large Supabase Table List
The speaker asked whether Supabase tables could be grouped by purpose, such as frontend or performance tables, instead of appearing in one long list. These units did not establish a confirmed solution.
- 0:52:06
An Agent Implementation and Review Loop
The speaker described a workflow in which an orchestrating agent passes along work and an auditor reviews it; if no issues are found, the task moves to done and the loop continues through the milestone.
- 0:53:13
Diagnosing a Supabase MCP Connection Failure
The speaker instructed the agent to investigate why Supabase MCP access was failing despite authorization, suggesting a quick search for the correct setup if it could not resolve the issue directly.
- 0:55:50
Different Supabase MCP Experiences in Codex and Claude
The speaker said Supabase MCP setup was failing for him in Codex despite being present in configuration, while access through Claude had been seamless. This was his experience at the time, not a general verdict on either tool.
- 0:56:18
An Opinion on Codex and Claude Effort Levels
Based on his personal experience, the speaker compared effort settings in Claude and Codex. He felt Claude’s levels had a more noticeable effect, while the distinction seemed less meaningful to him in Codex.
- 0:57:44
Discovering the Supabase Plugin in Codex
After struggling with MCP setup, the speaker noticed and installed a Supabase plugin in Codex. The agent then appeared to create a task list and work as requested, leading him to speculate that plugins may be Codex’s way of packaging MCP integrations; he did not establish that interpretation as fact.
- 0:58:27
Automatically Breaking Work into a Task List
The speaker praised an agent for creating and executing a task list on its own, contrasting this with his experience using another agent in the command center. He believed this decomposition could accelerate progress relative to the initial estimate, but these units did not establish milestone completion.
- 0:59:56
Using Agent Access to Review Table Organization
The speaker proposed using the agent’s Supabase access to analyze the tables and suggest a better organization, presenting it as a possible next step rather than confirmed completed work.
- 1:02:17
Reviewing Dependencies for the First Top-of-Funnel Phase
The speaker sets the session goal as advancing the first top-of-funnel phase and reviewing its tasks for dependencies on foundational work that may need to come first.
- 1:02:22
Organizing Supabase Tables into Logical Groups
The speaker discusses the growing difficulty of navigating many tables and proposes a clear map and categories. He then asks whether logical groupings would remain internal documentation or could be represented visibly in Supabase, while also considering ownership and physical organization.
- 1:06:25
Lessons from Git Branch and Worktree Conflicts and UI Recovery
The speaker recounts how accumulated branches, worktrees, and merges—followed by an agent-led cleanup without enough oversight—damaged the main version and restored an older UI. The agent later recovered parts of the work, leading the speaker to favor fewer branches and a better understanding of branch and worktree management.
- 1:11:28
Organizing database tables with schemas
Explores using a schema to group a large set of backend tables more clearly, starting by inspecting the current structure to understand it better.
- 1:14:46
Using a preserved browser state to help recover a website UI
Explains how an unrefreshed incognito tab preserved a visible version of a site after the current version broke, allowing reference screenshots and the latest code change to help guide recovery.
- 1:17:58
An impression of the Claude Code experience
The speaker compares moving to Claude Code with driving an electric car after being used to gasoline cars: a distinctly different experience even beside familiar, capable tools. This is a personal analogy, not a performance test.
- 1:19:30
Implementing GA4 and consolidating analytics tools
Discusses setting up GA4 and connecting it with other measurement tools in a unified metrics area, using a very concise step-by-step agent prompt and screenshots. It also presents an assistant-enabled browser that can see the current page as useful for navigation and account-setup tasks.
- 1:24:56
Lessons from recovering work scattered across Git branches and worktrees
Describes how accumulated, unmerged changes across multiple branches and worktrees led to a difficult, section-by-section recovery. The stated takeaway is that failing to merge and push to the main branch was a rookie mistake that became a practical workflow lesson.
- 1:29:48
An automated workflow for building a Shopify UI with Polaris
Describes work on a Polaris web component within a milestone whose sequential tasks are reviewed and approved by an automated auditor loop. After completion, the next step is to inspect how the application looks inside Shopify and iteratively build toward the reference design.
- 1:33:37
Using Screenshots as Context in Codex
Luqman explored supplying a screenshot to explain the current state and continuing the workflow in Codex, describing the approach as easy and convenient.
- 1:34:31
Organizing Environment Variables as a Project Grows
He explained that environment keys and variables accumulate as a project grows, making careful organization important when maintaining them across deployment services such as Railway and Cloudflare.
Got a question about something in this stream?
Ask in the group