[MCP Clients] execute_task requires long timeout (≥600s) - documentation should prominently warn users

[Jwd-gity]

Problem

OpenSpace MCP tool openspace::execute_task can run for several minutes, yet many MCP clients (including Hermes Agent) enforce a default 120-second timeout on MCP tool calls. This causes legitimate long-running tasks to be incorrectly terminated and reported as failures.

Specific Example - Hermes Agent:

• Default MCP tool call timeout: 120 seconds
• execute_task may need 5-10+ minutes for complex tasks
• Result: Hermes reports timeout while OpenSpace continues processing → orphaned tasks

Reproduction (Hermes)

1. Configure OpenSpace MCP in Hermes
2. Execute a complex task via openspace::execute_task
3. Hermes times out after 120s, OpenSpace keeps running

Recommended Fix

Add a prominent warning in OpenSpace docs:

⚠️ execute_task may run for 5-10 minutes or longer. MCP clients must set a timeout of at least 600 seconds (10 minutes).

Hermes Configuration Example

mcp_servers:
  harshal-mcp-proxy:
    url: http://localhost:8765/mcp
# Call execute_task with explicit timeout:
# timeoutMs: 600000

Impact

Affects all MCP clients with conservative timeout settings. Leads to silent failures and poor UX.

---

This issue was automatically reported via Hermes Agent MCP integration.