Claude Code deletes your session history after about thirty days.
Susan keeps what those sessions decided — searchable, for your whole team, and readable by every agent on it.
npm i -g @arkone_ai/susanDecisions survive the transcript
The CLI distils each session into what was actually decided and pushes it here. Searchable months later, by anyone on the team, long after Claude Code has deleted the conversation it came from.
Your agents read the team's context
Connect over MCP and every session starts knowing the conventions, the ownership and the constraints somebody wrote down once — instead of inferring them from the code and getting them wrong.
Skills, shared and vouched for
Write a skill once and every agent on the team has it. Publish it and other organisations can install it — with install counts and verdicts from teams who actually ran it, not stars.
Plans two people can see
A plan in Claude Code lives inside one session and dies with it. A shared one means the second engineer does not start the work the first is halfway through.
Three commands
npm i -g @arkone_ai/susan export SUSAN_HUB_TOKEN=sk_susan_… # from Settings, once you have signed in susan push # everything this session decided
Then point your agents at it:
claude mcp add --transport http susan https://susan.network/api/mcp \ --header "Authorization: Bearer sk_susan_…"
No billing yet, no SSO, and no rate limiting. The CLI is MIT and open; what your team pushes is private to your organisation.