the starter guide · step 10 of 13
your repo in innernet
for anyone who writes code. one command gives a repo its own memory, every commit keeps it current, and a teammate who clones it gets it too.
skip this step if you don't write code. if you do, it's the part of innernet that works hardest for you while asking the least.
one command, inside the repo
open a terminal in your project's folder and run:
npx innerneton top of connecting your tools (step 3), inside a repo it:
- offers to start a memory for it. say yes. netti reads the readme and the file tree and designs the pages that fit this project.
- adds a small file,
.innernet/connection.json, with the memory's name and nothing secret. commit it. it's how the link travels with the code. - installs two quiet git hooks: one after every commit, one after every pull. they send the commit messages to your memory in the background and can never slow git down.
- writes a short note into
AGENTS.md, so claude code, codex, cursor, copilot and gemini all know where the memory is and how to use it.
then just work
that's the setup. from here, every commit is noted, netti folds the notes into the project's pages, and the next session in any tool starts from the current picture, including why the last few things changed.
one habit makes it much better: write commit messages that say why, not only what. switch deposits to stripe so hosts get paid at check-in teaches your memory something. fix stuff doesn't.
see the repo
open the project on your dashboard and press the repo chip.
- each lane is a branch; each dot is a commit.
- hover a lane to light it up.
- click it for its commits, the whole change, and its pull request.
no repo linked yet? the view asks for your repo's https link. paste it, please, and it draws itself. activity → repos lists every repo you've tied to a memory.
the timeline chip, next to it, shows the same commits as they landed in memory, beside everything else that changed it.
end a session well
before you close for the day, ask your coding ai to leave a handoff. it writes down what got done and what's next. tomorrow, in any tool, where did we leave off? picks it up.
working with a team
because .innernet/connection.json is committed, a teammate who clones the repo and runs npx innernet reads the same memory, if you've shared it with them. if their account can't open it yet, the command tells them to ask you, instead of quietly starting a second memory.
ci and scripts
make a key under connections → api and anything can add to the memory:
INNERNET_API_KEY=innernet_… npx innernet capture "deployed v2.4 to production" -s ciyour keys
keys are for scripts only. never paste one into a chat.
everything here in technical detail: repo memory and the npx command.
next: the chat window.
the starter guide, every step
one memory, every ai tool.
innernet is memory your ai tools read live: every project, every decision, every skill, in one place, connected once. this guide is one of the parts worth learning; the rest you'll pick up by using it.