Facing issues while trying to use Atlassian MCP servers confluence tool in crew.ai agent and task

General
1

When I issue the crew Kickoff it looks like the corect tool gets invoked but soem how the pageID is not being passed as an input param to the tools interface. Any idea what I may be missing

:rocket: Crew: crew
└── :clipboard: Task: c6efeebe-63de-4834-bc54-0cd01334a424
Status: Executing Task…
└── :robot: Agent: Confluence Content Retriever
Status: In Progress

Agent: Confluence Content Retriever

Task:

OBJECTIVE:
Fetch the specified Confluence page and create a clear, concise summary of its contents.

PAGE TO FETCH:
**- Page ID: {1402839643}**

REQUIREMENTS:
1. Retrieve the full content of the specified Confluence page using the page_id and space_key
2. Create a concise summary (300-500 words) highlighting the main points
3. Extract key concepts and terminology
4. Note any important technical details or specifications
5. Format the output in a clear, readable structure

Keep the summary factual and focused on the most important information from the page.

:robot: Agent: Confluence Content Retriever
Status: In Progress
└── :brain: Thinking…
:robot: Agent: Confluence Content Retriever
Status: In Progress
:robot: Agent: Confluence Content Retriever
Status: In Progress
args ()
kwargs {‘page_id’: ‘’, ‘title’: ‘’, ‘space_key’: ‘’, ‘include_metadata’: True, ‘convert_to_markdown’: True}

Agent: Confluence Content Retriever

Thought: Action: confluence_get_page

Using tool: confluence_get_page

**## Tool Input: **
“{"properties": {"page_id": "1402839643", "include_metadata": "true", "convert_to_markdown": "false"}}”

Tool Output:

Error calling tool ‘get_page’
:robot: Agent: Confluence Content Retriever
Status: In Progress
└── :brain: Thinking…
:robot: Agent: Confluence Content Retriever
Status: In Progress

Logs from the actual MCP server indicates that pageID is being passed as Null

2

Hey @rrao, first off, welcome to our community!

As for your question, you’ll definitely need to give us a bit more context to have a better shot at getting help.

It looks like your Agent is understanding that the page id should be passed as page_id, but it seems you actually need it to be pageID. So, could you please show us how you’re setting up your MCP tools? That way, others can try to reproduce the error on their end and help figure out what’s going on.

3

Thx for the quick response. Looks like the filtering logic I had put to skip the Jira tools was causing the issue. Now I am just passing the entire tools and this seems to be working

from crewai_tools import MCPServerAdapter
from crewai import Agent, Task, Crew, LLM

server_params = {"url": "http://mcp-server:9000/sse"}

#confluence_tools = None
#with MCPServerAdapter(server_params) as all_tools:
    # Filter to only get Confluence tools
    #confluence_tools = [tool for tool in all_tools if tool.name.startswith('confluence_')]
#print(f"Available Confluence tools: {[tool.name for tool in confluence_tools]}")

confluence_tools = MCPServerAdapter(server_params)

confluence_fetcher = Agent(
    role="Confluence Content Retriever",
    goal="Fetch and deliver content from specified Confluence pages",
    backstory="""I'm a straightforward content retriever that connects to Atlassian MCP server 
    to fetch Confluence page content. I don't analyze or interpret the content - I simply retrieve 
    it accurately and deliver it in a clean format.""",
    **tools=confluence_tools.tools**,
    verbose=False,
    memory=False,  # Simple fetching doesn't require memory
    llm=LLM(
        model="openai/aiwb/llama-8b",
        temperature=0.1,  # Low temperature for consistent, predictable responses
        max_tokens=1000,
        extra_headers={
            "X-TFY-METADATA": '{"tfy_log_request":"true"}'
        }
    )
)

Below is the Actual output of a run that is able to retrieve the confluence page

🚀 Crew: crew
└── 📋 Task: c368f288-795f-4dbd-a394-8e77c94fbee2
       Status: Executing Task...
    └── 🤖 Agent: Confluence Content Retriever
            Status: In Progress
# Agent: Confluence Content Retriever
## Task: 
    OBJECTIVE:
    Fetch the specified Confluence page and create a clear, concise summary of its contents.

    PAGE TO FETCH:
    - Page ID: {975512419}

    REQUIREMENTS:
    1. Retrieve the full content of the specified Confluence page using the page_id and space_key
    2. Create a concise summary (300-500 words) highlighting the main points
    3. Extract key concepts and terminology
    4. Note any important technical details or specifications
    5. Format the output in a clear, readable structure

    Keep the summary factual and focused on the most important information from the page.
    
🤖 Agent: Confluence Content Retriever
    Status: In Progress
└── 🧠 Thinking...
🤖 Agent: Confluence Content Retriever
    Status: In Progress
🤖 Agent: Confluence Content Retriever
    Status: In Progress
args ()
kwargs {'page_id': '975512419', 'title': '', 'space_key': '', 'include_metadata': True, 'convert_to_markdown': True}


# Agent: Confluence Content Retriever
## Thought: Thought: To fetch the specified Confluence page and create a summary of its contents, I will first use the `confluence_get_page` tool to retrieve the page content using the provided page ID.
## Using tool: confluence_get_page
## Tool Input: 
"{\"page_id\": \"975512419\", \"include_metadata\": true, \"convert_to_markdown\": true}"
## Tool Output: 
{
  "metadata": {
    "id": "975512419",
    "title": "AI-Workbench on MKS (HLD/LLD)",
    "type": "page",
    "created": "",
    "updated": "",
    "url": "https://confluence.mastercard.int/pages/viewpage.action?pageId=975512419",
    "space": {
      "key": "ORBT",
      "name": "ONE Review Boards: Standard Governance"
    },
......
......