The Bug of add_history_to_context

General
1

Hi，This issue is very urgent.When I set add_history_to_context=True while creating the agent, this caused all the questions I had asked—displayed in the chat history—to revert to the content of my first question after I refreshed the chat page.

This is the normal chat page before refreshing：

image
image400×374 11 KB

This is the page after I refreshed it：

image
image377×357 9.19 KB

After my investigation, the issue lies with this API: Get:/sessions/{session_id}/runs — all the run_input fields it returns have turned into “Hi”.

image
image774×569 18.6 KB

When I set add_history_to_context=False, these issues do not occur.

The Agent UI also has this error.
This issue is very urgent for me. As you know, context memory is indispensable in chats.

2

Hi @LBWhite55, thanks for reaching out and supporting Agno. I’ve shared this with the team, we’re working through all requests one by one and will get back to you soon.If it’s urgent, please let us know. We appreciate your patience!

3

Yes, I’m quite in a hurry.

4

Hi @LBWhite55 ,

I’ve tried to replicate your issue, but haven’t been able to.
Could you let me know which version on the SDK you’re running so I can better support?

Also, if you’re not on latest, please make sure to update, as this issue might’ve been addressed already behind the scenes.

Thanks!

Ruan

5

agno is 2.0.6
this is my code：

pg_db_for_sltt = PostgresDb(
    db_url = db_url,
    session_table = "agno_sltt_session",
    memory_table = "agno_sltt_memory",
    knowledge_table = "agno_sltt_knowledge",
    id = "agent_sltt")

def get_sltt_agent():

    sltt_product = Milvus(
        uri = f"http://{config.MILVUS_HOST}:{config.MILVUS_PORT}",
        collection = "sltt_goods",
        embedder = embedder
    )

    sltt_product_knowledge = Knowledge(
        vector_db=sltt_product,
        contents_db=pg_db_for_sltt)

    # 创建AI助手
    assistant = Agent(
        name="sltt_Assistant",
        model=OpenAILike(
            id="Qwen3-Coder",
            api_key="...",
            base_url="https://www.sophnet.com/api/open-apis/v1",
        ),
        add_history_to_context=True,
        add_datetime_to_context=True,
        db=pg_db_for_sltt,
        knowledge=sltt_product_knowledge,
        use_json_mode=True,
    )
    return assistant

sltt_agent = get_sltt_agent()

agent_os = AgentOS(
    os_id = "my_agent",
    agents = [jfr_agent,sltt_agent],
    teams=[team],
)
6

Hey @LBWhite55 ! Could you confirm if this is still an issue? Could you try updating to the latest version and confirm if this is still an issue?