web
You’re offline. This is a read only version of the page.
close
Skip to main content

Announcements

News and Announcements icon
Community site session details

Community site session details

Session Id :
Power Platform Community / Forums / Copilot Studio / Copilot Studio + Asana...
Copilot Studio
Suggested Answer

Copilot Studio + Asana MCP: create_tasks fails with “expected array, received object” for tasks

(0) ShareShare
ReportReport
Posted on by

Hi all — I’m testing an integration between Asana’s V2 MCP server and a Copilot Studio agent. I created an Asana MCP app, connected it to Copilot Studio with OAuth, and the server appears reachable. Asana documents that the V2 MCP server uses OAuth, and Copilot Studio documents that MCP tool inputs/outputs come from the connected MCP server. 

The main problem is with task creation, not auth. When the agent tries to call create_tasks, I get this error:

MCP error -32602: Input validation error: Invalid arguments for tool create_tasks:
[
  {
    "code": "invalid_type",
    "expected": "array",
    "received": "object",
    "path": ["tasks"],
    "message": "Expected array, received object"
  }
]
 

In the Copilot Studio test pane, the tool input shows tasks(table) as required, which matches the validation error. The agent seems to be sending tasks as an object instead of an array/table inside the expected payload.

What’s confusing is that the model reasoning sometimes explicitly says it knows tasks should be an array, but the final call still fails validation. That makes this feel more like a request-shaping / tool-binding issue than an OAuth or endpoint issue. Copilot Studio says MCP tool schemas are surfaced from the MCP server, so I’m wondering whether anyone has seen a mismatch between the displayed tool schema and the actual payload being sent.

A few questions:


  • Has anyone successfully used Copilot Studio with Asana’s MCP create_tasks tool?

  • If so, what exact request shape is expected for create_tasks in this environment?

  • Has anyone seen Copilot Studio repeatedly pass tasks incorrectly even when the tool input shows tasks(table)?

  • Did you solve this through agent instructions, tool selection, or something else?

  •  

Any help would be appreciated — especially from anyone who has gotten write actions working reliably with the Asana MCP server in Copilot Studio.

I have the same question (0)
  • Suggested answer
    RaghavMishra Profile Picture
    185 on at

    Hi there! Great question — this is a nuanced MCP schema-binding issue.

    What's happening

    Copilot Studio surfaces MCP tool schemas directly from the connected MCP server (Asana's V2 MCP). When the tool schema declares tasks as a table (array), Copilot Studio should pass it as a JSON array. However, the generative orchestration layer sometimes serialises a single-item construct as a JSON object instead of a single-element array, which triggers Asana's -32602 validation error.

    What Microsoft Learn says about MCP in Copilot Studio

    According to the Extend your agent with Model Context Protocol documentation, tool inputs and outputs are shaped by the schema advertised by the MCP server, and Copilot Studio passes tool arguments based on that schema type.

    Suggested workarounds to try

    1. Add explicit instructions in your agent's system prompt — e.g., "When calling create_tasks, always pass tasks as a JSON array, even for a single task." Per Microsoft Learn, agent instructions can guide how the model shapes tool calls.
    2. Wrap in a Power Automate flow — Create an agent flow that accepts individual task fields as parameters, then constructs the array internally before calling the Asana API. This gives full control over the payload shape.
    3. Check the MCP server's JSON schema — Confirm the schema declares tasks with "type": "array" and test whether wrapping in an array via instructions resolves it.

    ⚠️ The precise request shape required by Asana's V2 MCP server is vendor-specific. For definitive schema documentation, refer to Asana's MCP server docs directly.

    References

    Found this helpful? Please mark ✅ "Does this answer your question?" so others searching for the same issue can find it quickly. A 👍 on "Was this reply helpful?" or a ♥ Like is also much appreciated!

    Raghav MishraLinkedIn | PowerAI Labs

Under review

Thank you for your reply! To ensure a great experience for everyone, your content is awaiting approval by our Community Managers. Please check back later.

Helpful resources

Quick Links

Season of Sharing Community Challenge Launch!

Jump in, show your community spirit, and win prizes!

Kudos to our 2025 Community Spotlight Honorees

Expanding mentorship, skilling, and AI innovation

Congratulations to the May Top 10 Community Leaders!

These are the community rock stars!

Leaderboard > Copilot Studio

#1
Valantis Profile Picture

Valantis 249

#2
Romain The Low-Code Bearded Bear Profile Picture

Romain The Low-Code... 180 Super User 2026 Season 1

#3
Vish WR Profile Picture

Vish WR 153

Last 30 days Overall leaderboard