Ramble the architecture. Your agent reads the memo.
Your best technical thinking happens away from the keyboard — and by the time you’re back at the terminal, the context is gone. Your AI coding agent could have used that.
Think out loud on the commute
Talk through the refactor: the options, the constraint you keep forgetting, the decision.
Dooli structures the decision
A decision memo or spec — options, tradeoffs, the call — stored as a real document.
Your agent picks it up via MCP
Dooli’s MCP server exposes your notes to Claude Code, Codex, or any MCP-capable agent. “Read my note on the auth refactor. Implement option B.”
What it looks like in practice
You say
“auth refactor — option A keep sessions, option B go full JWT, um, B wins because of the mobile client, but rotate refresh tokens, don’t let the agent touch the legacy SSO path…”
Dooli delivers · Decision memo
A memo your agent can act on: option B (JWT) selected with the mobile-client rationale, refresh-token rotation required, legacy SSO explicitly marked off-limits.
You say
“idea for the CLI — subcommand structure like git, config in the home dir, ship init, run, and doctor first, doctor checks the env and prints fixes…”
Dooli delivers · Website spec / feature spec
A feature spec: git-style subcommand layout, config location, a v1 scope of init / run / doctor, with doctor’s behavior defined.
Give your agent the context you spoke, not typed.
Three finished notes, free. Your notes stay on your device.
Try Dooli free