Bring your feeds into your team's memory.

Point Kythene at an RSS, Atom or JSON feed and it brings what's there into memory, kept up to date, in the project and tags you choose. Your team's AI recalls it alongside everything else it knows, and the owner is told when a feed stops working. Feeds are all it reads today.

Start free Read the docs →

for a person

Set it up once, then leave it

An owner adds a feed in Settings: its address, a name, and optionally the project and tags its items should carry. Kythene reads the feed before it saves, so one it can't read isn't added.

Kept up to date: It syncs about every hour. An entry that changes in the feed updates its memory in place, and Sync now reads the source straight away.
Told when it stops working: When a source starts failing, the owners get a line in their inbox, and your alert rules deliver it. Once for the failure, once more when it recovers - not every hour in between.
See what happened: Each integration shows its last run: how many items were added, updated, unchanged or skipped, and why any were left out. A failing one shows the error as the source gave it.
Pause it, or take it away: Disable one to stop syncing and keep everything. Remove it and what it brought in stays as ordinary memory.
Owners only: Setting these up is an owner's job. Everyone else just finds the results in memory.

for their AI

Recalled like anything else

There's no separate place to look. Each entry becomes a memory, so it comes back from the same recall call as what your team wrote by hand - in the flow of the agent's own work, with nobody pasting a link into a prompt.

One memory per entry: In the project and tags you chose, so it's found alongside the work it's about.
Marked as coming from a feed: Each memory carries a source:rss tag and names the site it came from, so an agent can tell what a person wrote from what a feed delivered.
What you delete stays deleted: Delete an item and it isn't brought back on the next sync.
Any MCP client: Claude, Cursor, Codex, Copilot and the rest - your team need not all use the same one. Connect yours →

An example

The release notes your team keeps meaning to read.

Say a framework you depend on publishes its release notes as a feed. Here's what an owner does, and what everyone gets.

Add the feed

Paste its address, name it, and send its items to your platform project with a dependencies tag.

It stays current

Each new release lands as a memory within the hour. If the feed stops working, the owner hears about it.

Agents use it

Ask an agent to plan an upgrade and it recalls what changed, next to what your team decided last time.

What about my other tools?

Feeds are the only source today, and only public ones. Kythene isn't a way to call other tools through one endpoint, and it isn't an automation platform: this is about getting what other systems know into what your team's AI remembers. What's available is what's on this page.

Where it fits

One part of the whole.

Kythene is where a team and its AI work together: what's in progress, what's been settled, what the AI applies, and now what the team's other systems have to say. Integrations feed the same memory the rest of it reads.

Every capability, by group →

What does it cost?

Pricing is by plan, and the pricing page is where the plans are set out.

See pricing →

The rest of it

Publish the work, agree what's true, bring in the rest.

artifacts

What other systems know

Feeds brought into memory and kept current. You're here.

Give your team's AI a feed to read.

Sign in, connect your AI, and add the first feed. Your agents are free, and so is anyone you invite as a guest to read and comment.

Start free Read the docs See pricing