lumbridge-code/docs/ROADMAP.md

Roadmap

Phase 0 — research and decisions

• Capture upstream snapshots, licenses, relevant modules, and architectural notes.
• Benchmark UI candidates with six animated terminal panes.
• Prove OpenSSH and tailscale ssh stdio transports against a disposable remote runtime, including disconnect and replay.
• Spike PTY correctness and process-tree cleanup on macOS, Ubuntu, and Omarchy.
• Connect the official ACP Rust SDK to two contrasting harnesses.
• Audit actual usage/quota surfaces for every launch provider.
• Prove Buzz SDK message, attachment, cursor, reconnect, and duplicate-delivery behavior against a disposable local relay.
• Decide terminal core, UI stack, IPC transport, and schema evolution policy.
• Define the Lumbridge Harness process boundary, trace-consent model, provider- neutral endpoint profile, and upstream-fork maintenance policy.

Exit: recorded decisions with working spikes and measured results.

Phase 1 — terminal workspace alpha

• Durable runtime, local IPC, workspaces, tabs, splits, shells, scrollback.
• Saved SSH/Tailscale hosts and durable remote panes through a per-user remote runtime; reconnect from a second Lumbridge installation.
• Session restore, crash recovery, command palette, keybindings, notifications.
• Per-pane Terminal, Browser, Tools, Context, Goal, and Review surface tabs, plus an inert decision shelf that can render typed suggestions without executing.
• Repository/worktree creation and basic Git status/diff.
• Optional Buzz collaboration pane, workspace channel binding, and confirmed redacted pane-image sharing.
• macOS, Ubuntu, and Omarchy development packages.

Exit: Lumbridge is worth using as a terminal multiplexer without AI features.

Phase 2 — agent alpha

• Generic harness manifest and PTY adapter.
• ACP client with capability negotiation and permission rendering.
• Codex, Claude Code, DeepSeek Harness, Pi, and Hermes adapters.
• Agent status, unread/needs-input state, resume, and transcript navigation.
• Separately versioned Lumbridge Harness with local/self-hosted OpenAI-compatible model support, prompt refinement, trace projections, and capability-gated workspace plans.

Exit: daily multi-agent work is reliable and observable.

Phase 3 — providers and usage

• Subscription profile discovery without credential extraction.
• BYOK profiles for OpenAI, Anthropic, Gemini, Groq, Cerebras, and DeepSeek.
• OS credential storage, redaction, provider/model catalog.
• Usage ledger, reset windows, burn charts, forecasts, export, and footer.

Exit: usage displays are useful, auditable, and honest about uncertainty.

Phase 4 — public beta

• Goose, Gemini CLI, OpenClaw, and OpenCode adapters.
• Signed/notarized macOS releases, .deb, AppImage, and Arch package.
• Delta updater with signed manifests and rollback.
• Accessibility, performance, soak, migration, and recovery qualification.