High-level milestones for the AgentStore project. Students implement features through Kanban tickets.
- Directory structure and placeholder files
- 10 agent manifests
- 8 tool manifests
- Mock datasets
- Documentation (README, architecture, user stories)
- Backend and frontend placeholders
- Agent listing API (
GET /agents) - Agent detail API (
GET /agents/{id}) - Tool listing API (
GET /tools) - Simulated agent run endpoint (
POST /agents/{id}/run) - Tool-call trace endpoint (
GET /agents/{id}/traces) - Rating/review endpoints
- JSON file storage layer
- Agent browse page with cards
- Agent detail page
- Tool browse page
- Run agent simulation UI
- Tool-call trace viewer
- Rating/review UI
- Agent registry loader
- Tool registry loader
- Mock agent runner
- 3+ complete tool-call traces
- Trending agents algorithm (mock data)
- Rating aggregation script
- Review sentiment placeholder
- Usage analytics dashboard (notebook)
- Remix/fork agent flow
- Publish new agent listing
- Version history display
- Agent search and filtering
- Category browsing
- Tool detail pages
- Agent-to-agent calling
- Agent App bundles
- Recommendation system
- Agent reliability dashboard
- Agent security scanner (permission audit)
- Agent leaderboards
- Real MCP server integration
- Database migration (SQLite/Postgres)
- User authentication
- Deployment pipeline
- Real external tool integrations
- Real payments
- Production authentication
- Real Gmail/Calendar/GitHub permissions
- Enterprise admin dashboard
- Live deployment
See docs/product-requirements.md for MVP boundaries.