This page is unfinished. Content and design are still in progress.
A single agent walks one path through a search space. A swarm searches many paths at once, coordinating parallel work while every agent stays aware of changes in the shared repository.
Intelligence is, at its core, a search problem: chess engines beat humans because they search the space faster and know where to search. The model supplies the intuition about which branches are worth trying, while the swarm explores many of them in parallel.
Spawn two or more agents in the same repo, and they will automatically be managed by the server to allow native collaboration. When agent A edits a file that agent B has read (code shifting under its feet), the server notifies agent B. Agent B can ignore it if it is not relevant, or it can check the diff to make sure that it doesn't conflict. Each agent has messaging abilities, capable of DMing just one agent, broadcasting to all other agents hosted by the server, or just agents working in that repo. This allows you to spawn multiple sessions in the same repo, and have all conflicts automatically resolved.
Jcode's server maintains the shared state that makes parallel work practical. Agents can direct-message one collaborator, broadcast to their spawned subtree, assign structured task-graph nodes, and wait for typed completion artifacts instead of relying on unstructured chat. File-read tracking tells an agent when code it has seen changes underneath it, while ownership and completion state remain visible to the coordinator. The result is parallel exploration without pretending the workers are isolated: each agent can move quickly, see the changes that matter, and reconcile overlapping work before it becomes a merge conflict.