Files
wehub-resource-sync 98e40dac97
CLI Smoke Test / smoke-test-linux (20) (push) Waiting to run
CLI Smoke Test / smoke-test-linux (24) (push) Waiting to run
CLI Smoke Test / smoke-test-windows (20) (push) Waiting to run
CLI Smoke Test / smoke-test-windows (24) (push) Waiting to run
Expo App TypeScript typecheck / typecheck (push) Waiting to run
chore: import upstream snapshot with attribution
2026-07-13 12:40:49 +08:00

4.2 KiB

Happy Roadmap

Next Up

  • Start using as the daily development driver
  • Contributing guidelines — priority: bugs > ui touchups > new features > refactors > core refactors (sync engine, rpc, server changes). Get notified about all issue activity on github — just start using inbox?
  • Start tweeting about changes with images / videos

Table Stakes (catch conductor)

  • Small UX touchups - too many to list

  • Bundled distribution (needed to make my own daily driver)

  • Forking a session - for example a session with triaging -> fork into multiple where we fix the specific groups of issues

  • File preview / editing in session — see layout-core.md

  • Better diff viewer — see layout-core.md

Navigation Bugs

Back navigation is broken across the app in several places:

  • Logout → restore from key doesn't pop enough screens (also errors out)
  • General back navigation inconsistency across flows

Workspaces & Checkouts

Missing the concept of workspace (aka project) that spans multiple machines, and checkout as a daemon-managed entity.

  • Workspace = a logical project that can span multiple machines (e.g. my laptop + cloud dev box both working on the same repo)
  • Checkout = a specific working copy on a specific machine, managed by the daemon
  • Currently we have machines and paths but no first-class workspace grouping
  • Daemon should manage checkout lifecycle: create worktree, switch branch, clean up stale checkouts
  • Right panel context (changes, files) is per-checkout, but workspace groups checkouts across machines

[hard]

  • Attachments in composer / in agent output [hard, encrypted attachments, extra storage - needs design]
  • Terminal embedded in app

Underlying Assistant Upkeep

  • Bug fixes, especially with session lifecycle management
  • Crons / scheduled agents
  • Migrate to most up to date vendor sdks
  • Better flags support

[nice to have]

  • Memory viewing / editing

  • Slash commands

  • Codex review, other commands

  • Tighter MCP / tool ecosystem hooks

  • Keep up as vendors ship new features

  • Cleaner protocol + unit tests

Viral / Cool

  • Multi-agent dispatch
    • Fan-out N agents across machines
    • Agents dispatching agents (you just watch)
    • Orchestration UI (progress, results, cost)
  • Software factories / maintenance factory
    • Repeatable agent pipelines
    • Own repo as first customer — self-maintaining
  • Voice
    • Dispatch agents by talking
    • Voice as the control layer

Talk to Users & Community

  • Reach out to 5 users directly
  • Read app store / google play reviews
  • In-app surveys / feedback chat
  • Contribution guidelines + PR template
  • Post about latest version
  • Engage with open PRs / community contributions

Growth

  • Semi-autonomous posting
    • Nudge-tweet after each ship
    • Watch git activity → draft posts
  • Semi-automated engagement
    • Find relevant people / conversations
    • Draft replies, human approves
  • Twitter / HN / socials presence

Session / Project Management

  • Reorder / prioritize sessions in sidebar

Customization

  • UI self-customization ("change X" → happy obliges)
  • Custom widgets per session / project
  • Widgets on mobile (iOS/Android) + desktop

Push Notification Routing

Current state: server stores bare Expo push tokens per account with no device metadata. All registered devices get all notifications — no routing intelligence.

  • Smart routing: use presence/activity signals (last active device, which device has the session open) to route notifications to the right device instead of blasting all
    • Server already has presence data (sessionCache, machine connection state) — likely enough to make good routing decisions
    • Suppress notification on the device that originated the action
    • Prefer the device the user is currently active on
  • Web push notifications: currently missing entirely — add service worker + web push registration so browser sessions get notifications too
  • Device metadata on token registration: store platform (ios/android/web), device name, last active timestamp alongside the push token

Better Machine Management

  • Auth transferring between devices

Integrations (external services)

  • remote machine ecosystem
    • exe.dev — tutorials, outreach
    • sprites — same
  • Linear
  • GitHub — PR reviews