​
Overview

The session context manager creates a new session, binds it to the current async/thread context, and automatically ends it when the context exits. This is the recommended pattern for self-contained workflows.

​
Syntax

with lai.session(
  session_name: Optional[str] = None,
  session_id: Optional[str] = None,
  api_key: Optional[str] = None,
  agent_id: Optional[str] = None,
  task: Optional[str] = None,
  providers: Optional[List[ProviderType]] = [],
  production_monitoring: Optional[bool] = False,
  mass_sim_id: Optional[str] = None,
  experiment_id: Optional[str] = None,
  rubrics: Optional[list] = None,
  tags: Optional[list] = None,
  masking_function = None,
  auto_end: Optional[bool] = True,
):
  ...

​
Parameters

  • session_name (string, optional): Display name for the session.
  • session_id (string, optional): Provide to resume under this ID; otherwise a new one is created.
  • api_key, agent_id (string, optional): Credentials; fall back to LUCIDIC_API_KEY and LUCIDIC_AGENT_ID.
  • task, rubrics, tags (optional): Session metadata.
  • providers (list, optional): Instrumentation targets like "openai", "anthropic", etc.
  • masking_function (callable, optional): Mask sensitive data before sending.
  • auto_end (bool, optional): Ignored inside this context; session always ends on exit.

​
Returns

None. The active session is bound to context for the duration of the with block.

​
Examples

import lucidicai as lai
from openai import OpenAI

with lai.session(session_name="Process Order", providers=["openai"]):
    client = OpenAI()
    resp = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[{"role": "user", "content": "Summarize the order"}],
    )

​
Notes

  • Spans are stamped with the active lucidic.session_id at start for correct attribution.
  • Prefer this for request-scoped work or jobs that should clean up automatically.

​
See Also