Roadmap / tracking: remaining milestones

Tracking issue for the remaining planned work. Grouped under the Roadmap milestone.

Shipped this cycle (context)

• M9a agent survival — re-spawn + adapter resume (#50)
• M9b full split-layout restore (#51)
• VT grid + content-based attention (#52)
• M7d remote TLS + token auth + TOFU (#53)
• Cleanup batch — dead-code removal, flake fixes, TLS-aware warning (#54)

Open

• #55 — M9c: PTY-host true zero-disruption agent survival. The survival capstone: agents (+ scrollback + live processes) survive a daemon restart with no re-spawn, via a persistent PTY-host + SCM_RIGHTS fd-passing. Design exists (docs/superpowers/specs/2026-08-02-clowder-m9-agent-survival-design.md §M9c). Big — its own spec → plan → SDD cycle.
• #56 — M8: Linux support. Broaden beyond macOS. The Rust daemon is likely portable; the SwiftUI/libghostty frontend is the hard part. Needs a scoping brainstorm first; cheapest early win is verifying the daemon runs on Linux drivable from the macOS app over the M7d remote transport.
• #57 — M7d follow-ups: remote TLS/auth polish. Deferred minors (token-in-logs, all-or-nothing rotation) + larger design-deferred items (mTLS, QUIC, pinned-pairing UX, Keychain).

Suggested order

M9c and M7d follow-ups are self-contained and can go anytime. M8 is the largest and should start with its own decomposition brainstorm. No hard dependencies between the three.

Start by reading the linked issues #55, #56, and #57, then review docs/superpowers/specs/2026-08-02-clowder-m9-agent-survival-design.md §M9c. This issue is a roadmap tracker rather than a self-contained change; it is done when the remaining work has been decomposed into actionable, separately scoped issues.