Skip to content

Blog

Notes, tips, and updates from the Fresh Jots team.

· 7 min read

AI assistant over MCP to connect Fresh Jots to — just talk to your notes

Every "write to Fresh Jots" guide assumes someone writes the integration — you, or your agent. The Model Context Protocol removes even that step: connect Fresh Jots once and your assistant reads and writes your notes by just asking. Here's how to turn on the Pro-tier MCP connector in Claude (browser), Cursor and VS Code (editor config), or Claude Code (console), the eleven tools you get, and the one prompt that makes your notebook your AI's session log.

Read more
· 23 min read

`/save-to-freshjots <title> :: <body>` — one running engineering journal, one titled append

Most AI session captures spin up a new note each time, burning your free-tier slots in ten sessions flat. /save-to-freshjots is the third shape: a single slash command that takes `<title> :: <body>` and appends each titled thought to one growing engineering journal at freshjots.com via the by-filename/append endpoint, plus a durable local backup written before every upload — your journal in two places, one growing note instead of a folder of orphans.

Read more
· 16 min read

Outbound webhooks — the user manual

Lock a note to append-only, paste a URL into its Settings, and every append turns into one POST. Three formats, one feature: signed JSON for your own server, native chat for Slack, native chat for Discord — no polling, no glue code. The full field-by-field reference, plus a working Sinatra receiver for signature verification. Pro & Team, included in the 14-day trial.

Read more
· 9 min read

Two easy steps to set up everything, and get you going

A Pro or Team subscription gives you an API token — but on its own it does nothing, and your AI agent will happily invent a Fresh Jots endpoint and fail. Two steps, once: persist the token in your shell profile, paste one installer line. After that, in any new terminal, "write this to my Fresh Jots" just works — through a real local CLI, never a guessed URL.

Read more
· 9 min read

Hook, slash command, skill, agent: who actually runs your AI?

Hook, slash command, skill, agent — four words Claude Code users treat as interchangeable. They aren't. One question sorts them all: who executes it, and does it run with your environment? Only one of the four can touch your token — here's which, and why it matters.

Read more