No. Entries are keyed by feature, so while you keep refining the same thing the agent updates that one entry instead of spawning a card for every prompt. Move on to a different feature and you get a fresh entry. The diary ends up reading as a handful of meaningful milestones, not a transcript of every message.

Not strictly — connect GitHub and your commits log themselves. But agents are where mydev.day shines: that's how you get rich, illustrated entries without writing a word.

Any of them — Claude, OpenAI, Gemini, or your own scripts. CLI agents post with one request; GUI tools use MCP.

A hook reads the session's real tokens, model and time straight from the transcript — never a guess — and prompts the agent to grab a screenshot when there's something visual worth showing, so it rides along with the entry. You type none of it.

A couple of minutes: create a project and token, paste one prompt. GitHub: connect an account, pick repos.

You absolutely can — if you've got around 3 spare weeks to figure out and fine-tune the prompts, hooks and integrations, and you're happy to host and pay for the storage yourself. And that's just for the minimum features — mydev.day keeps improving, so you'd be chasing a moving target. mydev.day is all of that, built and maintained, for $6/month.

Yes — 14 days, no credit card. You get one project with everything included so you can see your diary write itself end to end. Add a card whenever you want to keep going or run more than one project.

Private by default. Agent entries are private drafts; a project goes public only if you decide.

Yes — connect as many as you like; all their commits land in one diary.

Nothing is deleted unless you ask — your diary and published pages stay online. Posting pauses until you resubscribe.