tool invoke error: Instance <Account at 0x7f094a2ed7c0> is not bound toa Session; attribute refresh operation cannot proceed

[xiyanluo]

Self Checks

• I have read the Contributing Guide and Language Policy.
• This is only for bug report, if you would like to ask a question, please head to Discussions.
• I have searched for existing issues search for existing issues, including closed ones.
• I confirm that I am using English to submit this report, otherwise it will be closed.
• 【中文用户 & Non English User】请使用英语提交，否则会被关闭 ：）
• Please do not modify this template :) and fill in all the required fields.

Dify version

1.8.0

Cloud or Self Hosted

Self Hosted (Source)

Steps to reproduce

Publish the workflow as a tool, and then use an agent to execute the error message. There is no problem with independent workflow execution。
I have searched for other posts with similar issues, but they only mentioned the problem with the account object. How can I make the necessary changes? Can't this variable type be modified in the workflow?

response
f"getDouYinvideo": "tool invoke error: Instance <Account at 0x7f094a2ed7c0> is not bound toa Session; attribute refresh operation cannot proceed (Background on this error at:https://sqlalche.me/e/20/bhk3)"}

✔️ Expected Behavior

Call the tool normally by the agent

❌ Actual Behavior

No response

[dosubot]

This error happens because the Account object is being accessed outside of an active SQLAlchemy session—common when agent-invoked tools pass ORM objects across async boundaries or after the session is closed. Direct workflow execution keeps objects session-bound, but agent-tool invocation often serializes/deserializes data, so only IDs should be passed and ORM objects must be re-fetched inside a fresh session context.

To fix this, always pass the Account ID (not the object itself) between workflow nodes, agents, and tools, and then re-query the Account inside a session where you need it:

from sqlalchemy.orm import Session
with Session(db.engine) as session:
    account = session.query(Account).get(account_id)
    # use account safely here

If you must use an ORM object after the session is closed, re-attach it with db.session.merge(account) before accessing its attributes. Avoid passing ORM objects between threads, async calls, or outside the function where the session is active—convert them to dicts or pass only IDs if needed for later use.

Recent Dify updates recommend this pattern and have refactored session management to use explicit context managers, which helps prevent this error. If you still see the error in 1.8.0, check your workflow/tool code for places where ORM objects are passed instead of IDs and update accordingly. Upgrading to the latest version is also recommended, as fixes for this issue have been merged recently (details, details, details).

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}    

[xiyanluo]

This error happens because the Account object is being accessed outside of an active SQLAlchemy session—common when agent-invoked tools pass ORM objects across async boundaries or after the session is closed. Direct workflow execution keeps objects session-bound, but agent-tool invocation often serializes/deserializes data, so only IDs should be passed and ORM objects must be re-fetched inside a fresh session context.

To fix this, always pass the Account ID (not the object itself) between workflow nodes, agents, and tools, and then re-query the Account inside a session where you need it:

from sqlalchemy.orm import Session
with Session(db.engine) as session:
account = session.query(Account).get(account_id)
# use account safely here
If you must use an ORM object after the session is closed, re-attach it with db.session.merge(account) before accessing its attributes. Avoid passing ORM objects between threads, async calls, or outside the function where the session is active—convert them to dicts or pass only IDs if needed for later use.

Recent Dify updates recommend this pattern and have refactored session management to use explicit context managers, which helps prevent this error. If you still see the error in 1.8.0, check your workflow/tool code for places where ORM objects are passed instead of IDs and update accordingly. Upgrading to the latest version is also recommended, as fixes for this issue have been merged recently (details, details, details).

To reply, just mention @dosu.

How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other    

The latest version of 1.8.0 has been updated to the latest version, but it still doesn't work. The node code has been changed. Where should I change it specifically?

[dosubot]

To fix the session binding error, you need to update your workflow/tool node code so that you only pass the account_id (not the Account object) between nodes, agents, and tools. Wherever you need to use the Account, re-query it inside a fresh SQLAlchemy session. This pattern is used throughout Dify's workflow and tool node logic—for example, in ToolNode, parameters are generated from primitive values (like IDs), and objects are fetched inside session contexts.

Before (problematic):

# Passing ORM object between nodes/tools
def my_node(account: Account):
    # ... use account directly ...

After (correct):

# Pass only account_id between nodes/tools
def my_node(account_id: str):
    from sqlalchemy.orm import Session
    with Session(db.engine) as session:
        account = session.query(Account).get(account_id)
        # ... use account safely here ...

Or, if you use a repository/helper:

def get_account_by_id(account_id: str):
    with Session(db.engine) as session:
        return session.query(Account).get(account_id)

Where to change:

• In your workflow node and tool code, update all function signatures, parameters, and variable passing so that only IDs (not ORM objects) are passed between steps.
• In each node/tool where you need the Account, open a session and re-query by ID as shown above.
• Do not cache or hold onto ORM objects between async calls, threads, or outside the session scope.

Relevant files to check:

• api/core/workflow/nodes/tool/ (ToolNode and related nodes)
• api/core/repositories/ (database access helpers)
• Any custom workflow/tool node code you wrote that passes Account objects

If you still see the error after making these changes, try restarting your Docker container to reset session state for legacy tools (reference).

Test by verifying that only IDs are passed between nodes/tools and that no DetachedInstanceError occurs when accessing Account attributes.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}    

[QuantumGhost]

@xiyanluo Thank you for reporting this issue.

To help us investigate and resolve this problem effectively, could you please provide:

1. Full stack trace - This will help us pinpoint the exact location where the error occurs
2. Workflow DSL export - You can export this from the workflow editor to help us reproduce the issue
3. Detailed reproduction steps - A step-by-step guide on how to trigger this behavior

[xjw00654]

This issue seems to be the problem of the Async Workflow Execution, I guess.

for chatflow error, change the code from api/libs/helper.py:L33 function extract_tenant_id from the original impl. to return 'xxxxxx-xxx-xx' # Your tenant id. It could be alternative solution to temperarily solve it.

for the agent (workflow as tool) error, i am oning to find the solution, update later ....

------ sep ------

for agent (workflow as tool scenario) the alternative solution would be:

at api/services/account_service.py, ~148L; add 2 lines:

if naive_utc_now() - account.last_active_at > timedelta(minutes=10):
    account.last_active_at = naive_utc_now()
    db.session.commit()
account.id   # <<<< access the variable `account.id`
db.session.expunge(account) # <<<< use `sqlalchemy.expunge` to make the varibale decouple with above session
return cast(Account, account)

Hope it would be helpful :D

[crazywoola]

This issue seems to be the problem of the Async Workflow Execution, I guess.

for chatflow error, change the code from api/libs/helper.py:L33 function extract_tenant_id from the original impl. to return 'xxxxxx-xxx-xx' # Your tenant id. It could be alternative solution to temperarily solve it.

for the agent (workflow as tool) error, i am oning to find the solution, update later ....

------ sep ------

for agent (workflow as tool scenario) the alternative solution would be:

at api/services/account_service.py, ~148L; add 2 lines:

if naive_utc_now() - account.last_active_at > timedelta(minutes=10):
account.last_active_at = naive_utc_now()
db.session.commit()
account.id # <<<< access the variable account.id
db.session.expunge(account) # <<<< use sqlalchemy.expunge to make the varibale decouple with above session
return cast(Account, account)
Hope it would be helpful :D

@xjw00654 Hello whether you can submit a PR to fix this? It seems you have already addressed the issue.

[QuantumGhost]

Hi, @xjw00654,

the detached Tenant issue should be fixed in #24789. The detached Account error appears to be a separate issue that may need additional handling.