--- title: My current agentic coding workflow | daily.dev description: A developer walks through his personal agentic coding workflow using a self-built CLI tool called Nebula. He covers three main patterns: working directly off... canonical: https://daily.dev/posts/my-current-agentic-coding-workflow-39devzqky twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: My current agentic coding workflow | daily.dev og:description: A developer walks through his personal agentic coding workflow using a self-built CLI tool called Nebula. He covers three main patterns: working directly off... og:url: https://daily.dev/posts/my-current-agentic-coding-workflow-39devzqky og:image: https://api.daily.dev/og/posts/39DevZqKY.png og:image:alt: My current agentic coding workflow og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # My current agentic coding workflow **[Web Dev Cody](https://daily.dev/sources/webdevcody)** · 13 min read · 2 upvotes · 0 comments ## Summary A developer walks through his personal agentic coding workflow using a self-built CLI tool called Nebula. He covers three main patterns: working directly off main for side projects, using isolated git worktrees plus pull requests for team projects, and running agents on remote VPS/sandboxes for safety and uptime. He shares model choices (Claude Code for implementation/planning, Codex for review, Grok 4.5/4.6 for quick questions), always using the highest effort/model tier despite cost, and discusses security concerns around letting agents run unsupervised on work laptops versus isolated, credential-scoped environments. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### Why should I run Claude Code or Codex agents on a VPS instead of my local laptop? Running coding agents on an isolated VPS or in a Docker container limits security risk since agents can read, write, and delete files and could act maliciously if something goes wrong, even though Claude Code's sandboxing is generally reliable. A VPS also has far more uptime than a laptop, letting long agent sessions (30 minutes to a couple hours) run overnight without being interrupted by a dead battery or lost internet connection. _daily.dev surfaces workflow write-ups like this for developers weighing local versus sandboxed agent setups._ ### Should I use git worktrees or work directly on main when running coding agents? Working off main is fine for side projects or internal tools where code review isn't required, since it lets you commit and push directly once an agent finishes a task. For team projects requiring review, creating a separate git worktree isolates the agent's changes so you can inspect a git diff, open a pull request, and let teammates review before merging. _developers weighing worktree-based review flows against direct-to-main commits can compare approaches on daily.dev._ ### Which AI models are best for different steps of an agentic coding workflow like implementation, review, and quick questions? One practical split is using Claude (Fable 5 or Opus 5) at extra-high effort for initial implementation and planning, Codex at its highest setting for reviewing code changes, and Grok 4.5 or 4.6 for quickly answering questions about the changed code because of their speed. Sticking with the highest-effort model throughout, rather than switching to faster or cheaper models, avoids the extra cognitive overhead of fixing buggy or incomplete output later. _daily.dev helps developers track how others split model choice across planning, review, and quick Q&A tasks._ ## Similar posts on daily.dev - [Agentic coding finally clicked when I wired it to a kanban board](https://daily.dev/posts/agentic-coding-finally-clicked-when-i-wired-it-to-a-kanban-board-8oagsdeet) · XDA Developers · 1 upvotes · 0 comments - [Claude Code as an Autonomous Agent: Advanced Workflows \(2026\)](https://daily.dev/posts/claude-code-as-an-autonomous-agent-advanced-workflows-2026--4gjoc6yvx) · SitePoint · 1 upvotes · 0 comments - [5 Agentic Coding Tips & Tricks](https://daily.dev/posts/5-agentic-coding-tips-tricks-48zhdqcpx) · Machine Learning Mastery · 2 upvotes · 0 comments --- Tags: [#ai-agents](https://daily.dev/tags/ai-agents), [#ai-coding](https://daily.dev/tags/ai-coding), [#claude-code](https://daily.dev/tags/claude-code) [View this post on daily.dev](https://daily.dev/posts/my-current-agentic-coding-workflow-39devzqky) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"My current agentic coding workflow","url":"https://daily.dev/posts/my-current-agentic-coding-workflow-39devzqky","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/my-current-agentic-coding-workflow-39devzqky"},"datePublished":"2026-08-23T14:54:15.843Z","dateModified":"2026-09-13T21:59:50.136Z","description":"A developer walks through his personal agentic coding workflow using a self-built CLI tool called Nebula. He covers three main patterns: working directly off...","image":"https://i.ytimg.com/vi/xLQgOzK1adE/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/xLQgOzK1adE/sddefault.jpg","isAccessibleForFree":true,"articleSection":"Web Dev Cody","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Web Dev Cody","logo":"https://media.daily.dev/image/upload/s--9ioIrvxf--/f_auto/v1723789220/logos/webdevcody","url":"https://daily.dev/sources/webdevcody"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/my-current-agentic-coding-workflow-39devzqky","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":2},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-agents,ai-coding,claude-code","timeRequired":"PT13M","video":{"@type":"VideoObject","name":"My current agentic coding workflow","description":"A developer walks through his personal agentic coding workflow using a self-built CLI tool called Nebula. He covers three main patterns: working directly off...","thumbnailUrl":"https://i.ytimg.com/vi/xLQgOzK1adE/sddefault.jpg","uploadDate":"2026-08-23T14:54:15.843Z","duration":"PT13M","url":"https://api.daily.dev/r/39DevZqKY","embedUrl":"https://www.youtube.com/embed/xLQgOzK1adE"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Web Dev Cody","item":"https://daily.dev/sources/webdevcody"},{"@type":"ListItem","position":3,"name":"My current agentic coding workflow"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/my-current-agentic-coding-workflow-39devzqky#faq","mainEntity":[{"@type":"Question","name":"Why should I run Claude Code or Codex agents on a VPS instead of my local laptop?","acceptedAnswer":{"@type":"Answer","text":"Running coding agents on an isolated VPS or in a Docker container limits security risk since agents can read, write, and delete files and could act maliciously if something goes wrong, even though Claude Code's sandboxing is generally reliable. A VPS also has far more uptime than a laptop, letting long agent sessions (30 minutes to a couple hours) run overnight without being interrupted by a dead battery or lost internet connection. daily.dev surfaces workflow write-ups like this for developers weighing local versus sandboxed agent setups."}},{"@type":"Question","name":"Should I use git worktrees or work directly on main when running coding agents?","acceptedAnswer":{"@type":"Answer","text":"Working off main is fine for side projects or internal tools where code review isn't required, since it lets you commit and push directly once an agent finishes a task. For team projects requiring review, creating a separate git worktree isolates the agent's changes so you can inspect a git diff, open a pull request, and let teammates review before merging. developers weighing worktree-based review flows against direct-to-main commits can compare approaches on daily.dev."}},{"@type":"Question","name":"Which AI models are best for different steps of an agentic coding workflow like implementation, review, and quick questions?","acceptedAnswer":{"@type":"Answer","text":"One practical split is using Claude (Fable 5 or Opus 5) at extra-high effort for initial implementation and planning, Codex at its highest setting for reviewing code changes, and Grok 4.5 or 4.6 for quickly answering questions about the changed code because of their speed. Sticking with the highest-effort model throughout, rather than switching to faster or cheaper models, avoids the extra cognitive overhead of fixing buggy or incomplete output later. daily.dev helps developers track how others split model choice across planning, review, and quick Q&A tasks."}}]} ```