We've added MCP server access so teams can connect Custory to external AI clients like Cursor, Claude, and other compatible agents using workspace-scoped credentials. This opens up a more direct way to use Custory from the AI environments many teams already work in, while keeping access tied to the right workspace. Set it up from [Manage Integrations](https://app.usecustory.com/manage-integrations) under **MCP Server**.
## What's New
**Workspace-scoped MCP keys.** MCP access is scoped to a workspace so teams can connect external AI tools without using broad, ambiguous access patterns. Keys can be created and revoked from inside the product, making it easy to manage access over time — especially when a client is no longer in use or should stop working immediately.
**Setup guidance built in.** Custory provides the practical setup details teams need, including ready-to-paste configuration snippets for supported MCP clients. This reduces the friction of getting connected and helps teams move from "this sounds useful" to "this is working" faster.
**AI workflow portability.** With MCP access in place, teams can work with Custory from the AI tools they already use for broader product and engineering workflows — letting those agents work with real Custory workspace context more directly.
## Why it matters
Many teams are already using external AI tools as part of how they plan, write, review, and execute work. MCP Server Access helps Custory plug into that reality more cleanly, so the product can participate in those workflows without forcing teams into a disconnected setup.
v0.1.17
MCP Server Access
Connect workspace-scoped MCP access so external AI clients can work with Custory more directly and securely.