Keep useful developer context across ChatGPT, Claude, Cursor and Terminal.

The problem is not storage

A good prompt can stay buried in a chat. A command can live in shell history on the wrong project. A code example can remain inside an old repository. A link can survive as a bare URL after the image and reason that made it useful are gone.

Notes, bookmarks and Gists all work when you already know where the item belongs. Working context appears earlier - while you are still deciding whether it deserves a folder, tag or document.

Capture, recognize, reuse

1. Capture deliberately. Save only the fragment that earned a place. Nothing is collected in the background.
2. Recognize before organizing. Keep the source, language, image, file name or thumbnail that lets you identify it in a second.
3. Reuse the original. Inspect without changing your clipboard, then copy the original payload back into the active editor, terminal or chat.

What belongs on a developer context shelf

• A prompt that produced a useful review or implementation plan.
• A CLI command with the flags you will not remember next month.
• A code block, API response or configuration fragment worth adapting.
• A documentation page, GitHub issue or post whose visual preview matters.
• A screenshot, PDF, video or file you need to bring into another tool.

A concrete cross-tool flow

1. Find a useful prompt in Claude and copy it.
2. Open Refetch with the gesture you chose in onboarding and save it.
3. Keep working in Cursor or Terminal.
4. Later, search by the words you remember and inspect the matching row.
5. Press Command-C and paste the original into the next task.

Why deliberate capture matters

Automatic memory has a real advantage: you never have to remember to save. It also creates a larger, noisier record of whatever crossed your screen or clipboard. Refetch chooses the opposite tradeoff. The small capture action is the signal that an item matters.

Where Refetch deliberately stops

Refetch is not cross-device sync, a team knowledge base, automatic AI memory or a passive clipboard manager. It is a native, local shelf for one developer's reusable working context on a Mac running macOS 26.4 or later.

If selected-text capture, platform support or automatic history changes the decision, compare Refetch with Copper, Cooper and other AI-work scratchpads.

Start with one real fragment

The 14-day trial is the full app. Save something you expect to need again, then judge Refetch when you actually retrieve it - not only when you first capture it.