Multi-agent AI laboratory with ASP.NET Core 8.0 backend and Flutter frontend. Implements CQRS architecture, OpenAPI contract-first API design. BACKEND: Agent management, conversations, executions with PostgreSQL + Ollama FRONTEND: Cross-platform UI with strict typing and Result-based error handling Co-Authored-By: Jean-Philippe Brule <jp@svrnty.io>
143 lines
4.8 KiB
Markdown
143 lines
4.8 KiB
Markdown
# API Changelog
|
|
|
|
This document tracks **breaking changes only** to the Codex API. Non-breaking additions and bug fixes are not included here.
|
|
|
|
## Format
|
|
Each entry should include:
|
|
- **Date**: When the change was introduced
|
|
- **Version**: API version (when versioning is implemented)
|
|
- **Endpoint**: Affected endpoint(s)
|
|
- **Change**: Description of what changed
|
|
- **Migration**: How to update client code
|
|
- **Reason**: Why the change was necessary
|
|
|
|
---
|
|
|
|
## [1.0.0-mvp] - 2025-10-26
|
|
|
|
### MVP Release - Frontend Ready
|
|
|
|
This is the initial MVP release with all core functionality complete and tested. The backend is **production-ready** for frontend integration.
|
|
|
|
#### Commands (6 endpoints)
|
|
- **POST /api/command/createAgent** - Create new AI agent with model configuration
|
|
- **POST /api/command/updateAgent** - Update existing agent settings
|
|
- **POST /api/command/deleteAgent** - Soft delete an agent
|
|
- **POST /api/command/createConversation** - Create conversation (returns Guid)
|
|
- **POST /api/command/startAgentExecution** - Start agent execution (returns Guid)
|
|
- **POST /api/command/completeAgentExecution** - Complete execution with results
|
|
|
|
#### Queries (4 endpoints)
|
|
- **POST /api/query/health** - Health check endpoint
|
|
- **POST /api/query/getAgent** - Get single agent by ID
|
|
- **POST /api/query/getAgentExecution** - Get execution details by ID
|
|
- **POST /api/query/getConversation** - Get conversation with messages and executions
|
|
|
|
#### List Endpoints (6 GET endpoints)
|
|
- **GET /api/agents** - List all agents (limit: 100 most recent)
|
|
- **GET /api/conversations** - List all conversations (limit: 100 most recent)
|
|
- **GET /api/executions** - List all executions (limit: 100 most recent)
|
|
- **GET /api/agents/{id}/conversations** - Get conversations for specific agent
|
|
- **GET /api/agents/{id}/executions** - Get execution history for specific agent
|
|
- **GET /api/executions/status/{status}** - Filter executions by status
|
|
|
|
#### Security & Infrastructure
|
|
- CORS configured for development (any localhost port) and production (configurable)
|
|
- Rate limiting (1000 requests/minute per client)
|
|
- Global exception handling middleware
|
|
- FluentValidation on all commands
|
|
- API key encryption for cloud providers (AES-256)
|
|
|
|
#### Documentation
|
|
- Complete API reference in `docs/COMPLETE-API-REFERENCE.md`
|
|
- Architecture documentation
|
|
- XML documentation on all commands/queries
|
|
- Enum reference for frontend integration
|
|
|
|
#### Database
|
|
- PostgreSQL with EF Core
|
|
- Full schema with migrations
|
|
- Soft delete support
|
|
- Proper indexing for performance
|
|
|
|
#### Design Decisions
|
|
- **Pragmatic over Perfect**: Simple GET endpoints instead of complex dynamic query infrastructure
|
|
- **MVP-First**: 100-item limits are sufficient for initial use case
|
|
- **No Pagination**: Can be added in v2 based on actual usage patterns
|
|
- **Client-Side Filtering**: Frontend can filter/sort small datasets efficiently
|
|
|
|
#### Known Limitations (Non-Blocking)
|
|
- Authentication not yet implemented (documented for v2)
|
|
- Swagger documentation only includes 5 endpoints (OpenHarbor limitation)
|
|
- All 16 endpoints are **functional and tested**
|
|
- Complete documentation provided in markdown format
|
|
- No real-time updates (WebSockets/SignalR planned for v2)
|
|
|
|
#### Next Steps
|
|
- Frontend team can start integration immediately
|
|
- Use `docs/COMPLETE-API-REFERENCE.md` as API contract
|
|
- Dynamic filtering/pagination can be added in v2 if needed
|
|
|
|
---
|
|
|
|
## [Unreleased]
|
|
|
|
### Future Enhancements
|
|
- JWT authentication and authorization
|
|
- Real-time execution updates via SignalR
|
|
- Advanced filtering and pagination (if usage patterns require it)
|
|
- API versioning strategy
|
|
- Distributed caching layer
|
|
|
|
---
|
|
|
|
## Example Entry Format
|
|
|
|
### 2025-01-15 - v2.0
|
|
|
|
#### POST /api/command/UpdateUser
|
|
**Change**: `username` field changed from optional to required
|
|
|
|
**Migration**:
|
|
```diff
|
|
{
|
|
- "username": null,
|
|
+ "username": "required_value",
|
|
"email": "user@example.com"
|
|
}
|
|
```
|
|
|
|
**Reason**: Data integrity requirements - all users must have unique usernames
|
|
|
|
---
|
|
|
|
## Guidelines for Maintaining This Log
|
|
|
|
### What qualifies as a breaking change?
|
|
- Removing or renaming endpoints
|
|
- Removing or renaming request/response properties
|
|
- Changing property types (string → number, nullable → required, etc.)
|
|
- Adding required fields to existing requests
|
|
- Changing endpoint HTTP methods
|
|
- Modifying authentication requirements
|
|
- Changing error response formats
|
|
|
|
### What is NOT a breaking change?
|
|
- Adding new optional fields
|
|
- Adding new endpoints
|
|
- Bug fixes that restore documented behavior
|
|
- Performance improvements
|
|
- Internal refactoring
|
|
|
|
### Process
|
|
1. Before making a breaking change, discuss with team
|
|
2. Update this CHANGELOG.md with the planned change
|
|
3. Notify frontend/API consumers with reasonable notice period
|
|
4. Implement the change
|
|
5. Re-export `openapi.json` via `export-openapi.sh`
|
|
6. Verify frontend teams have updated their clients
|
|
|
|
---
|
|
|
|
*Last updated: 2025-01-26*
|