Summaries let you store condensed versions of conversations and search across them. Use them to build session history, generate user profiles, or give agents quick access to past interaction highlights.
Store a Summary
from lyzr import Cognis
cog = Cognis()
result = cog.store_summary(
owner_id="user_alice",
session_id="sess_001",
content="Alice asked about Python decorators. We covered basic syntax, common patterns (@property, @staticmethod), and she practiced writing a caching decorator.",
messages_covered_count=12,
agent_id="tutor_bot", # optional
)
Parameters
| Parameter | Type | Required | Description |
|---|
owner_id | str | Yes | User/tenant identifier |
session_id | str | Yes | Session being summarized |
content | str | Yes | The summary text |
messages_covered_count | int | Yes | Number of messages this summary covers |
agent_id | str | No | Agent identifier |
Get Current Summary
Retrieve the latest summary for a session:
summary = cog.get_current_summary(
owner_id="user_alice",
session_id="sess_001",
)
print(summary.get("content"))
Search Summaries
Search across all archived summaries for a user:
results = cog.search_summaries(
owner_id="user_alice",
query="Python decorators",
limit=5,
)
for r in results.get("summaries", []):
print(f"[{r['session_id']}] {r['content'][:100]}...")
Parameters
| Parameter | Type | Required | Description |
|---|
owner_id | str | Yes | User/tenant identifier |
query | str | Yes | Semantic search query |
session_id | str | No | Filter to a specific session |
limit | int | No | Max results (default: 5) |
Async Variants
await cog.astore_summary(owner_id="user_alice", session_id="sess_001", content="...", messages_covered_count=12)
summary = await cog.aget_current_summary(owner_id="user_alice", session_id="sess_001")
results = await cog.asearch_summaries(owner_id="user_alice", query="decorators")
Summaries are a hosted-only feature. The open-source lyzr-cognis package does not support summaries.
