Behavior hints for clients (MCP 2025-03-26); hints, not guarantees.
Human-readable description of what the tool does.
Unique tool name.
Display name for clients (MCP 2025-06-18).
Validates, anonymizes and stores the feedback.
Tool arguments validated against the input schema
The stored feedback result with a text summary
MCP tool for collecting transformation feedback for AI learning. Stores feedback in the user's project (
<projectRoot>/.claude-feedback).Remarks
Record kinds:
transformation(default): outcome of a transform-code runanalysis: a read-only tool result (find-usages, search-codebase, ...) that had to be corrected via grep, build output, a script or manual readingPrivacy:
Storage:
ANGULAR_MODERNIZER_PLATFORM_ROOTboth(default): the project, plus the platform store when configuredproject/platform(platformneeds the variable)