Migrate a website from a CMS to raw code and Markdown
Lee Robinson used Cursor agents to move cursor.com off its CMS in three days for $260 in tokens, with subagents handling a broad refactor.
Done withCursor
- What they did
- Lee planned the migration with Opus 4.5 in Cursor. The agent suggested using the existing CMS API key to write scripts that export content, validate it, convert it to Markdown files in the repo, and upload images and videos to object storage. About 10 runs got it 80% of the way. He then ran an agent on each mismatched page, using browser screenshots to compare against production. Subagents handled the wider refactor in parallel.
- How it went
- It took three days and $260.32 in tokens, against his estimate of 1-2 weeks. That covered 344 agent requests and 67 commits (+43K/-322K lines). Builds got 2x faster. The last 20% took most of the time.
- Worth knowing
- Expect the final 20% (pages that don't match production) to take most of the effort. Per-page agent runs with screenshot comparison were how he closed that gap.
Try it yourself with Cursor
I want to move my website at [repo or folder] off [CMS name] and into plain code with Markdown files for content. First export the existing content from [CMS or export location], then use subagents to convert pages and posts into Markdown and update the templates and routes to read from those files. Finished means the site builds locally, the pages match the originals, and you give me a list of anything you could not migrate. Show me the plan before you delete any CMS code or integrations.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
1 more account of this
Source: leerob.com · Undated
Related use cases
Keep coding agents running autonomously for long stretches
Cursor's write-up on scaling long-running autonomous coding.
Build a personal research wiki from papers and articles
Cursor Agent writes and maintains interlinked markdown pages from user-supplied sources, summarizing and cross-referencing, with Obsidian used to browse.
Monitor tasks and open pull requests with long-running agents
A commenter runs a persistent script that watches tasks and uses Cursor to open pull requests, building a documentation summarizer and a data pipeline.
Build a C compiler with a team of parallel agents
Anthropic's engineering write-up on tasking Opus 4.6 agent teams with building a C compiler.
Turn customer feedback into tested pull requests
In OpenAI's own launch example, a Dot watches feedback for recurring requests, scopes small fixes, builds and tests them, and returns PRs with demo videos.
Build and launch a code-snippet sharing app
The poster built the Snippra app with Replit Agent, which handled UI copy, feature planning, debugging and naming while they guided it.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.