Future Vision
Homebrew Tap
Idea: brew install indigoai/tap/create-hq as an alternative to npx.
Why: Faster install, no Node.js dependency for the installer itself, native feel on macOS.
How: Create a Homebrew tap repo (indigoai-us/homebrew-tap) with a formula that downloads the compiled binary. Use pkg or bun compile to create a standalone executable from the TypeScript source.
Blocked by: v5 launch (need stable npm packages first).
Worker Marketplace
Idea: A registry where users can discover and install community-built workers.
How it could work:
hq modules add https://github.com/someuser/worker-email-digest# Clones the worker, adds to workers/, registers in modules.yamlWorkers are already just YAML + markdown files in git repos. The module system already supports adding remote repos. The marketplace would be a curated list (GitHub repo or web page) of verified workers.
Components:
- Worker registry (JSON/YAML hosted on GitHub)
- Submission process (PR to registry repo)
- Quality checks (required fields, valid YAML, description)
- Star/rating system (GitHub stars as proxy)
- Categories: coding, content, research, automation, data
Team Features
Idea: Shared HQ for teams — multiple people contributing workers, knowledge, and projects.
Challenges:
- Conflict resolution for simultaneous edits
- Permission model (who can edit what)
- Shared vs. personal workspace sections
- Real-time sync between team members
Possible approach:
- Shared S3 bucket with per-user prefixes for workspace/
- Shared prefix for knowledge/ and workers/
- Git-based collaboration for knowledge repos (already works)
- Real-time: WebSocket notifications when shared files change
Real-Time Collaboration
Idea: Multiple users editing the same file simultaneously (Google Docs-style).
This is hard. Requires:
- Operational Transform (OT) or CRDT for conflict-free editing
- WebSocket server for real-time updates
- Presence indicators (who’s viewing/editing)
- Cursor sharing
More realistic first step: File-level locking. If User A opens a file for editing, User B sees “locked by User A” and can view read-only.
Native Apps
Idea: Native iOS and Android apps instead of PWA.
Pros: Better performance, native notifications, background sync, biometric auth. Cons: Two codebases to maintain, App Store review process, higher development cost.
Verdict: Stay with PWA for now. Revisit if PWA limitations become blockers. React Native could bridge the gap if needed.
AI Model Flexibility
Idea: Support multiple AI providers, not just Claude Code.
Considerations:
- HQ workers currently depend on Claude Code’s slash command system
- Knowledge bases and CLAUDE.md are Claude-specific
- Supporting other models would require an abstraction layer
- Worker YAML could specify model preference
Near-term: Stay Claude-focused. The deep integration is a feature, not a limitation. Long-term: Abstract the agent interface so workers can run on different models.
Plugin System
Idea: Extend HQ with plugins for integrations (Slack, Linear, GitHub, Notion).
How it could work:
plugins: slack-notify: repo: https://github.com/indigoai-us/plugin-slack config: webhook_url: $SLACK_WEBHOOKPlugins would be lightweight adapters that hook into HQ events (task complete, sync conflict, checkpoint saved) and take actions (send Slack message, create Linear ticket, update Notion page).
CLI Improvements
Near-term CLI enhancements:
hq status— Dashboard view of HQ health (sync, workers, projects)hq doctor— Diagnose issues (missing deps, broken symlinks, stale locks)hq upgrade— Update HQ template to latest version (smart merge)hq export— Export HQ snapshot for backup/sharing- Tab completion for all commands (bash/zsh/fish)
Contributing Ideas
Have an idea? Open an issue on GitHub with:
- Problem: What’s missing or broken?
- Proposal: How should it work?
- Scope: Small fix or large feature?
We prioritize ideas that make HQ more useful for solo developers and small teams.