REC–0001ARCHIVED
Recueil
self-hosted webpage bookmarker and archiver
Recueil keeps a full copy of every page you save — the original, a clean readable version, and a search index — on hardware you control. Not a pile of links that quietly rot.
Bookmarks rot. Archives shouldn’t.
Pages get edited, paywalled, or taken down entirely. A saved link is a promise your browser doesn’t keep — the page you bookmarked and the page you find later are rarely the same one.
I ended up running a stitched-together stack of separate tools to cover that gap, each good at one piece and fighting me on the rest. Recueil is the tool I actually wanted: one purpose-built system for keeping what you save.
Capture happens in your browser tab
Not a headless crawler fetching pages cold. Whatever you can already see — logged in, past the paywall, scrolled open — is what gets saved.
- 1
Send a page
From the extension, your phone’s share sheet, a Shortcut, or the CLI.
- 2
Open it, once
The extension loads it in a real, already-authenticated tab.
- 3
Capture
A complete, self-contained snapshot — styles, images, and all.
- 4
Archived
Readable, searchable, and stored on your own hardware.
More than a copy of the page
- search
- Every capture is searchable by its full extracted text, not just its title.
- organize
- Tags and nested collections, organized your way.
- notes & links
- Keep notes on a page and link it to related saves.
- history
- Re-save a page later and keep every version, not just the latest.
Point an AI client at your library
Recueil runs a read-only MCP server. Connect a local MCP client and ask it what you saved about a topic last month, instead of digging through tags by hand.
MCP Server docs →Runs on hardware you control
Go and Postgres, deployed with Docker Compose. The whole capture path is outbound-only — no inbound ports required to keep the queue moving.