Summary
The workspace split is real, but the live runtime still depends on partially integrated paths and documented migration gaps. We should finish the extraction so crate boundaries match the actual runtime and there are fewer half-migrated seams.
Scope
- Identify and remove
TODO(integrate) / transitional ownership seams in the live runtime
- Reduce duplicated responsibility across
tui, cli, app-server, and shared crates
- Move source-of-truth logic out of monolith-era paths where split crates already exist
- Update architecture docs so they describe reality rather than the migration plan
Acceptance Criteria
- Fewer runtime-critical modules are owned only by
crates/tui because of temporary migration shortcuts
- Transitional seams in
features, streaming, and related runtime wiring are either integrated or removed
- Workspace docs reflect the real source-of-truth modules
cargo check and relevant tests pass after the refactor
Starting Points
docs/workspace_migration_status.md
docs/ARCHITECTURE.md
crates/tui/src/main.rs
crates/tui/src/features.rs
crates/tui/src/tui/streaming.rs
Summary
The workspace split is real, but the live runtime still depends on partially integrated paths and documented migration gaps. We should finish the extraction so crate boundaries match the actual runtime and there are fewer half-migrated seams.
Scope
TODO(integrate)/ transitional ownership seams in the live runtimetui,cli,app-server, and shared cratesAcceptance Criteria
crates/tuibecause of temporary migration shortcutsfeatures, streaming, and related runtime wiring are either integrated or removedcargo checkand relevant tests pass after the refactorStarting Points
docs/workspace_migration_status.mddocs/ARCHITECTURE.mdcrates/tui/src/main.rscrates/tui/src/features.rscrates/tui/src/tui/streaming.rs