Use-cases
Currently neither Claude.ai / Desktop nor OpenClaw support the MCP elicitation function. This means that while they can see the tools for creating no-code workspaces, they fail when asked to surface variable requests to the end user.
Attempted Solutions
MCP Elicitation does not seem to be widely supported at the moment. I have looked into options with both OpenClaw and Claude but have had no success.
Proposal
I would like to see an alternative where the LLM can query the no code modules' variable requirements first then submit them as it would for a normal workspace build, i.e. the required variables first then add the optional variables after the workspace has been created. By default, Claude falls back to the create_workspace + create_workspace_variable method - but can't link the workspace to the no-code module in this manner.
Use-cases
Currently neither Claude.ai / Desktop nor OpenClaw support the MCP elicitation function. This means that while they can see the tools for creating no-code workspaces, they fail when asked to surface variable requests to the end user.
Attempted Solutions
MCP Elicitation does not seem to be widely supported at the moment. I have looked into options with both OpenClaw and Claude but have had no success.
Proposal
I would like to see an alternative where the LLM can query the no code modules' variable requirements first then submit them as it would for a normal workspace build, i.e. the required variables first then add the optional variables after the workspace has been created. By default, Claude falls back to the
create_workspace+create_workspace_variablemethod - but can't link the workspace to the no-code module in this manner.