Generate large amounts of code with Manus
A user had Manus produce a lot of code quickly for software projects, but lost work a couple of times and had to add extra checkpointing.
Done withManus
- What they did
- The poster says they used Manus on one project for six months and spent close to 300,000 credits in the last few months, using it to generate a very large codebase quickly. They say Manus sometimes lost work, so they added a dual checkpoint system, and they also had a human dev shop audit the code while another LLM, GPT, coached or audited Manus when it got stuck.
- How it went
- They say the approach produced what they describe as about a million dollars of code and roughly a year of coding in under two months for less than $3,000. Their main shortfall was occasional lost work, which they say was mostly fixed by the extra checkpointing.
- Worth knowing
- Their practical warning is to add redundant checkpoints because they lost work a couple of times; they also recommend paying humans to audit along the way.
Try it yourself with Manus
Write the code for [describe the software project, language, and features]. Work in small pieces, and after each piece export the files and give me a checkpoint summary listing what exists and what is next so I never lose progress. Stop when all features run and the tests pass, and show me the full file list before you delete or overwrite anything.
Want help getting this working? Join the Vibe Coding Academy ↗ Run by Alex Finn, who also runs this site.
Source: Reddit · Undated
Related use cases
Compile journalist lists and search apartment listings
A reporter tested Manus on building a reporter list, searching NYC rentals and finding award candidates; results were thin at first, and a three-hour search yielded only three full profiles.
Generate slide decks from titles and a branded template
A reviewer had Manus find content for slide decks and produce a monthly marketing report in an uploaded PPTX template; editing tools were limited.
Produce competitor research reports and build a landing page
A team ran a multi-source competitor research report and a working landing page build through Manus, logging output and fixes needed.
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.
Click through apps to verify its own work
With computer use, Devin built and played a desktop game, ordered on Amazon, and tested its own work by clicking through the app.
Point a computer-using bot at any repo
A commenter says Grok Bot can use a computer and browser and be told to work on any given repo, unlike a repo-specific cloud agent.
Get 5 like this every morning
The best things people got an AI agent to do, each with the prompt to try it.