I’m getting following error running the L_4 notebook. The file path seems right.
Any suggestions?
“{"name": "support_tickets_data.csv", "description": "", "args_schema": {}, "return_direct": false, "verbose": false}”
Error: the Action Input is not a valid key, value dictionary.
I am having the exact same issue when running this lesson, Funny enough the problem seems to be isolated to the suggestion_engine agent:
“{"name": "support_tickets_data.csv", "description": "", "args_schema": {}, "return_direct": false, "verbose": false}”
Error: the Action Input is not a valid key, value dictionary.
The Reporting agent seems to use the tool fine and immediatly provides an expected output of what it found:
Thought: I need to access the support tickets data to generate the requested tables summarizing the key metrics and trends in the support data.
“{"name": "./support_tickets_data.csv", "args_schema": {}}”
ticket_id,customer_id,issue_type,issue_description,priority,date_submitted,response_time_minutes,resolution_time_minutes,satisfaction_rating,customer_comments,agent_id,resolved
T0001,C0511,API Issue,I’m pleased with how my issue was handled. Thanks!,High,2023-03-25,240,927,4,I’m pleased with how my issue was handled. Thanks!,A004,True
T0002,C0729,Login Issue,Excellent service! The agent went above and beyond.,Low,2023-04-06,223,534,5,The problem still persists. Not resolved yet.,A004,False… and so on.
Hey,
same here , is there any suggestion how to fix it ?
is the error output comes with you are running in course provided environment or your own local environment?
Also if possible can you share screenshot of the complete error you got
I need the screenshot of the complete error not a section.
The top header where it show ValueError and then explain the path from where error has generated.
can you take separate two screenshots if it is lengthy error log.
hi @openpnt is the tool output for all 3 agents same??
Hi ,
I’ve just run the delivered code for the test . 
I assume that it runs for all 3 agents in the background .
Hello @Anki337, @amut and @openpnt
So what I have understood the tool output error is because of the crewai tool version discrepancy.
What I would recommend first to click File==>Open, then requirement.txt which provides information about which version the codes assignment need to run.
So check once in course environment, Check if it matches with requirement.txt file version
if it matches and still gives the error output for agent, please let us know.
I am also sharing a GitHub link where people have encountered similar error, but the course instructor has responded he is looking into it.
I am also going to tag someone from staff, so this issue is addressed.
Thank you all.
P.S. please let me know about the version of python and crewai tool your notebook codes are running in the course provided environment with a screenshot for a better understanding for staff to address upon this further in case it doesn’t work.
Regards
DP
Hi @Deepti_Prasad ,
Here are the versions:
python : 3.11.9
crewai : 0.75.0
crewai-tools : 0.13.2
I 've realized that “Report Generator” agent can read the csv file content , while “Suggestion Agent” can not .
Tool input values look different for both tasks:
“{"name": "support_tickets_data.csv", "description": "", "args_schema": {}, "return_direct": false, "verbose": false}”
Error: the Action Input is not a valid key, value dictionary.
“{"name": "./support_tickets_data.csv", "args_schema": {}}”
ticket_id,customer_id,issue_type,issue_description,priority,date_submitted,response_time_minutes,resolution_time_minutes,satisfaction_rating,customer_comments,agent_id,resolved
I am sharing a screenshot of the requirement.txt version
make sure version matches, pip install the tool version requirement.txt mentions for crewai tool
Regards
DP
I have downgraded crewai_tools from 0.13.2 > 0.12.1 but that throws me another more sinister error. This time it is when creating the agents instead:
If I then --upgrade to 0.13.4 it works as it did in the beginning with version 0.13.2 that was installed when i first ran the notebook.
So on version 0.12.1 i get an attribute error and on 0.13.2=>0.13.4 the original error occurs.
When running newer versions, only the “Agent: Suggestion Engine” fails to use the csv_tool and cant read the csv, the other Agents such as " Agent: Report Generator" use it just fine so I dont imagine the error being due to package versions but something else. Otherwise all agents would likely provide the same error output in my opinion.
Are the any versions in between 0.12.1 and 0.13.2 that might work perhaps?
Thank you @Anki337 for trying.
Ok this error makes me now look at the other metadata file as it does point to the cvs tool issue.
Can you check the python version in the course environment once!!
Let me go through other metadata file, although I am doubting this version discrepancies between the agents and tool or probably when the codes were written it must have worked. Please give me sometime.
Just a heads up I have informed this issue to staff.
Regards
DP
I just had look at lab again, what I understood is that the filereadtool used to access the support.data.csv which has been recalled as CSV tool, is the one creating that error.
I looked through another py file as well as directory file which mentions the tool should give output about the search as what kind api issue was there and how it was handled.
So if this tool is giving an error output of invalid key, then the course creator has to look if the api key for crewai tool is still working or not.
So sorry to disappoint you right now and I would suggest wait for the staff to respond, so they coordinate with this issue.
I appreciate all your patience and reporting in working on this issue.
Regards
DP
Hey!
No problemo! It was a stunningly good course anyway and as it always is with sw, issues occur, so dont worry about that! Just glad we got to some sort of solution! Even if that might take a bit of time ! 
Thank you for reporting! We’ll be checking this with the team and will update you, here!
