| name | python-developer |
| description | Develops, modifies, and debugs Python services, LangChain chains, Vector Store connectors, and Streamlit dashboards within the Oncoflow application. |
Python Developer Skill
This skill provides comprehensive rules, code patterns, and reference architectures for managing Python development on the Oncoflow platform.
Key Technologies
- Python
>=3.13 (Check and strictly follow typing and typing extension patterns).
- Streamlit: Single-page and multi-page configurations using
st.Page and state synchronization via st.session_state.
- LangChain: Modern LangChain chains, custom agent wrappers,
ToolRetryMiddleware, ModelRetryMiddleware, and Pydantic validation.
- Milvus / Chroma: MMR retrieval using
max_marginal_relevance_search parameters (k, fetch_k, expr).
Rules & Patterns
1. LangChain Agent Integration
When modifying or instantiating an agent, always refer to the base template in agent.py. Utilize structured response structures based on Pydantic BaseModel:
from pydantic import BaseModel, Field
class CustomResponse(BaseModel):
clinical_summary: str = Field(description="Structured clinical description in French")
is_complete: bool = Field(description="True if record is complete")
2. Streamlit Dashboard Best Practices
- Never use heavy processes inside standard Streamlit loops without catching exceptions or using spinners (
st.spinner()).
- Keep date formatting robust and avoid comparing naive and aware datetimes. Use the utility:
def to_naive(dt: datetime | None) -> datetime | None:
if dt is None:
return None
if dt.tzinfo is not None:
return dt.astimezone(pytz.timezone("Europe/Paris")).replace(tzinfo=None)
return dt
3. Vector Database Connection Rules
- Do not build separate DB clients inside local search operations.
- Always retrieve context from standard readers passed through the LangChain runtime context:
reader: DocumentReader = runtime.context["reader"]
retrieved_docs = reader.vecdb.clientdb.max_marginal_relevance_search(
query, k=k, fetch_k=20, **kwargs
)
