Projects / Case study

prsist

One memory that Claude, Claude Code and Codex share. It turns sessions into short notes, files them by project, and hands the right ones to the next chat.

Role
Solo builder
Timeline
October 2026 to now
Built with
Python, FastAPI, FastMCP, Supabase, Fly.io
Status
Private alpha

The problem

I use Claude on my phone, Claude Code on three machines and Codex for reviews. None of them knew what I'd told the others. I was explaining the same project decisions again in every new chat.

Each tool's own memory only works inside that tool, and it keeps whole conversations, which is more than I want stored.

What I built

Hooks in Claude Code and Codex send each session to prsist as it happens. The chat apps connect as an MCP server. On the server, Claude Haiku distils the session into short notes and decides whether each one belongs to a project or applies everywhere. When it isn't sure, it asks me, and my answer becomes data for tuning.

New sessions start with the notes for their project. Trust is the point: prsist keeps notes, not conversations, and the design deletes every transcript within 24 hours.

What the distiller wrotePrompt v6, one session
  • The user plans to add a dashboard to prsist that lets them manually correct notes filed or kept global incorrectly.Project note, plan
    Expected
  • The user's Mac sleeps after 30 idle minutes, so long runs on it need caffeinate -i running in the background.Global note, fact
    Expected
  • In prsist, the user treats a wrong filing of a note into a project as a worse error than leaving the note global.Project note, decision
    Acceptable
  • In prsist, the user notices when a note they saved goes missing from the context shown to their AI assistants.Project note, fact
    Junk
  • The user's work machine is a Mac.Global note, fact
    Junk
  • ExpectedMatches a note I labelled by hand as one I'd want.
  • AcceptableNot in my labels, but the judge model says it's worth keeping.
  • JunkThe judge says it's obvious, too vague or not worth storing.

How I measure it

I labelled 39 real sessions by hand with the notes I'd want from each. A hash of the session id puts about a third of them in a test split that I never tune on. Each prompt version runs three times over the dev split, so I can see how much the score moves on its own. Claude Opus judges any note that isn't in my labels.

Two numbers matter. Capture is the share of my expected notes the distiller wrote. Junk is the share of what it wrote that the judge threw out.

Distiller prompt v5 vs v6, dev split
PromptSessionsCaptureJunkCost per eval
v51961.5% 58–6327.5% 27–28$0.94
v61761.5% 58–6522.8% 21–25$0.84
Three runs per version, range across runs. Distiller: Claude Haiku 5.5. Judge: Claude Opus 5.5. v6 ran on 17 sessions because I took out two where T3 Code was running another agent, which isn't the user talking.

v6 cut junk from 27.5% to 22.8% and kept capture where it was. A whole eval run costs less than a dollar, so I can afford to run one after every prompt change.

What's next