A larger product, platform, business domain, or technical system containing one or more related projects.
One logical codebase or repository within a system. This is the default memory isolation boundary.
One concrete local checkout or worktree of a project.
One Codex conversation lifecycle in a workspace.
A high-value structured memory item intended to help future sessions act faster or more correctly.
A structured continuation record written when a session pauses, checkpoints, or ends.
A compact continuation packet derived from prior memory during session bootstrap.
Another project in the same system that may be retrieved only under controlled policy.
A record that tracks imported data from secondary sources such as watchers or relay logs.
The local working copy from which the active Codex session is operating.
The logical repository or codebase associated with the current workspace.
The broader product grouping associated with the current project.
A record explicitly written by Codex through the memory tool flow.
A record or artifact imported from a local watcher or cache reader.
A record or artifact imported from a relay-side capture path.
A record synthesized during recovery after interruption.
Memory from the current workspace or current project.
Retrieval that includes a different project from the same system.
Retrieval that reaches outside the current system. This is never default behavior.
Durable, reusable, continuity-relevant information such as decisions, bug root causes, confirmed fixes, discoveries, constraints, preferences, and unresolved todos.