AGENTS.md - Adaapt AI Knowledge Base

Build/Test Commands

Run specific test category:

python run_tests.py --category unit|integration|security|e2e

Run single test:

python -m pytest tests/path/to/test.py::test_function -v

Start dev server:

python run_server.py

uvicorn app.main:app --reload

FastAPI app with async PostgreSQL, Redis, MinIO (S3-compatible), ChromaDB vector store
Database: SQLAlchemy models in
```
app/models/
```
(User, KnowledgeDomain, Document, DocumentChunk)
API structure:
```
/api/v1/
```
with auth, knowledge-base, ingestion, chat, agentic endpoints
Services: Business logic in
```
app/services/
```
(ingestion, embedding, llm, agentic_rag)
LangGraph: Agentic RAG workflows in
```
app/graphs/
```
Testing: Comprehensive RBAC test suite (unit, integration, security, e2e)

Imports: Use absolute imports from
```
app.
```
, group by standard/third-party/local
Models: SQLAlchemy with Mapped[] annotations, UUID primary keys, proper relationships
Error handling: Raise HTTPException with proper status codes, structured logging with structlog
Types: Full type hints with pydantic models for request/response schemas
Authentication: JWT-based with domain-based RBAC, dependency injection via
```
get_current_user
```
Async: Use async/await throughout, AsyncSessionLocal for database sessions

Run specific test category:

python run_tests.py --category unit|integration|security|e2e

Run single test:

python -m pytest tests/path/to/test.py::test_function -v

Start dev server:

python run_server.py

uvicorn app.main:app --reload

FastAPI app with async PostgreSQL, Redis, MinIO (S3-compatible), ChromaDB vector store
Database: SQLAlchemy models in
```
app/models/
```
(User, KnowledgeDomain, Document, DocumentChunk)
API structure:
```
/api/v1/
```
with auth, knowledge-base, ingestion, chat, agentic endpoints
Services: Business logic in
```
app/services/
```
(ingestion, embedding, llm, agentic_rag)
LangGraph: Agentic RAG workflows in
```
app/graphs/
```
Testing: Comprehensive RBAC test suite (unit, integration, security, e2e)

Imports: Use absolute imports from
```
app.
```
, group by standard/third-party/local
Models: SQLAlchemy with Mapped[] annotations, UUID primary keys, proper relationships
Error handling: Raise HTTPException with proper status codes, structured logging with structlog
Types: Full type hints with pydantic models for request/response schemas
Authentication: JWT-based with domain-based RBAC, dependency injection via
```
get_current_user
```
Async: Use async/await throughout, AsyncSessionLocal for database sessions