AI news · October 18, 2026
Concord now keeps several repositories in one task store
Concord 0.14.1 adds project tags to a shared store and works without concord setup; test it across two repos before centralizing agent messages.
- release
- v0.14.1
- machine-wide store
- 1
Concord's October 17 release adds global mode and project-scoped tasks. It can work in repositories where concord setup was never run, place tasks in one machine-wide store, and attach a project to each task so several repositories do not mix their work. The changelog also says Cursor can join global mode through MCP roots and a session hook. Signed-in users get Concord Cloud as the default store, with a shared general workspace for work outside a repository.
This is a small but practical change for people who run several coding agents on one machine. A shared task store can show who owns a task and where it belongs, but it also creates a new place where context can be mixed if the project tag is wrong. Test two repositories with different task names and inspect every read and write before using the cloud store for a team.
What you can do with it
Use two test repositories with separate task names. Create one task in each, ask an agent to inspect its project, and confirm that messages, files, and status stay with the right repo. Then try the Cursor connection and record setup time, context errors, and review work.
Our take
Shared task memory is useful when several agents need to coordinate, but the project name must remain visible at every step. Concord's global mode reduces setup work; it does not remove the need to check which repository a task belongs to. A two-repo test is enough to expose the main failure.
Links Try itRead the changelog
Source: Concord AI — Made With Models writes the brief; the reporting is theirs.