[DX] memwal_ask tool is undocumented and its behavior is unclear

[zazadra]

Problem

The memwal_ask tool is referenced in the MCP plugin and mentioned in community-built system prompts (e.g., for a 'Never Ask Twice' pattern), but there is no documentation explaining:

• How it differs from memwal_recall?
• Does it perform semantic search?
• Does it return a boolean, a relevance score, or the full blob content?
• What does an empty/no-match response look like?

Impact

Developers building system prompts that use memwal_ask to prevent redundant questions cannot reason about its behavior, leading to unpredictable agent responses.

Suggested Fix

Add documentation and an example request/response for memwal_ask in the README or MCP tool schema description. A concrete example showing the response shape when a match is found vs. not found would be sufficient.

[github-actions]

Thank you for opening this issue, a team member will review it shortly.

[ducnmm]

Thanks for the report — you're right that memwal_ask behavior isn't documented anywhere. The team has received this and is reviewing; we'll follow up here with docs / example request-response shapes.