ContextPassport carries your project — goals, decisions, progress, next steps — from one AI to the next. Watch the whole loop below, then try it yourself.
A real reproduction of the extension flow — click any step above to jump to it.
Every model switch, every rolled-over thread, every new morning — the same re-explaining.
Each of these is a step in the walkthrough above.
Short demos instead of screenshots. Each one plays when it scrolls into view.
Pasting a whole chat log wastes the next model's attention. The intelligence engine pulls out the six things that actually carry a project forward and drops everything else.
so basically we need the thing to keep working when I move to claude
yeah I already decided manifest v3, don't want a backend
I finished the chatgpt and gemini adapters yesterday
not sure what to charge for the founder tier honestly
anyway next up is the product hunt launch on friday
Captures used to blend into one pile, so handoffs came out thin and mixed. Each capture now carries a stable thread key built from the provider and the conversation URL — a new chat always becomes a new, distinctly named thread.
A research brief and a developer handoff need different shapes. Pick a mode and the same stored context is re-rendered for it — tuned per provider. Click through them here.
Goal: ship v1 to 100 founders
Decisions: MV3, no backend, local storage
Progress: 4 adapters shipped
Next: Product Hunt launch Friday
The pre-capture preview shows exactly what will land: turns found, turns already captured, and any integrity warning naming the adapter and selector responsible.
0–100, computed from what the passport is missing. Vague goal, undocumented decisions, stale blockers — each one shows up as a specific recommendation instead of a vibe.
Add a measurable success criterion to the goal
Two decisions have no recorded rationale
No blockers logged in 6 captures — still accurate?
Keep-last-N and max-age rules prune old captures automatically, pinned facets always survive. Export a whole workspace as one bundle and re-import with merge, dedupe or overwrite.
There is no backend to send anything to. Captures live in chrome.storage.local, the passport is rendered in your browser, and the network tab stays empty the whole time.
If one site changes its markup, only that provider degrades — the rest keep working, and the popup tells you which adapter tripped and why.
No account, no telemetry, no cloud copy. The demo below is what your network tab looks like during a capture.
| Compare | Free | Pro | Founder |
|---|---|---|---|
| Projects | 3 | Unlimited | Unlimited |
| Captures per project | 20 | Unlimited | Unlimited |
| AI providers | 4 | 4 | 4 + early access |
| Quick & Balanced modes | Yes | Yes | Yes |
| Developer Handoff | — | Yes | Yes |
| Research Handoff | — | Yes | Yes |
| Founder Handoff | — | — | Yes |
| Context Health Score | — | Yes | Yes |
| Project Memory | — | Yes | Yes |
| Recurring cost | $0 | $9 / mo | $0 after $39 |
A Chrome extension that moves your working context between ChatGPT, Claude, Gemini and Perplexity, so you never restart a project just because you switched tools.
It doesn't move the transcript. It extracts goals, decisions, constraints, progress, open questions and next actions, then re-renders them in the shape the receiving AI needs.
No. There is no server. Everything is stored in your browser on your device, and you can confirm zero outbound requests in DevTools → Network.
The extension reads the conversation visible on the page, structures it, and saves it locally. You see a preview of exactly what will be saved first.
Never. It writes the handoff into the composer and stops. You review it and press Send.
ChatGPT, Claude, Gemini and Perplexity at launch. Each provider is an isolated adapter, so if one site changes its markup only that provider degrades.
Yes, both are toggles in the popup's Privacy panel, and there's a one-click delete for all local data.
Everything in Pro forever with no recurring charge, plus Founder Handoff mode and early access to new providers.
Install it once. The next time you switch models, your context comes with you.