Summary
UnusedSymbolResult exposes UnusedBucket (likely_unused_private / maybe_unused_nonpublic / public_or_exported_no_refs / reflection_or_config_suspect), UnusedConfidence, and UnusedReason fields on the model. The classification logic exists (closed #105), but the MCP / --json response shape does not document the taxonomy — clients see opaque enum strings with no per-bucket explanation. Adding a --by-bucket aggregator and an explicit reason-taxonomy schema would make the output far more usable for automation.
This is distinct from #1403 (false-positive on same-file private), #1405 (BCL dominance in hotspots), and closed #105 (implementation of classification). C231 is about exposing the existing taxonomy through a documented schema.
Where
src/CodeIndex/Models/QueryResults.cs:39-44 (UnusedSymbolResult fields)src/CodeIndex/Mcp/McpToolHandlers.cs:1444-1447 (response serialization)
Suggested approach
(1) Document the bucket enum values in USER_GUIDE.md and the MCP tool description. (2) Add an aggregator field summary: { by_bucket: {...}, by_confidence: {...} } to the JSON envelope. (3) Add --exclude-tests, --bucket=… filter flags so callers can drill into one classification at a time without re-querying.
Summary
UnusedSymbolResultexposesUnusedBucket(likely_unused_private / maybe_unused_nonpublic / public_or_exported_no_refs / reflection_or_config_suspect),UnusedConfidence, andUnusedReasonfields on the model. The classification logic exists (closed #105), but the MCP /--jsonresponse shape does not document the taxonomy — clients see opaque enum strings with no per-bucket explanation. Adding a--by-bucketaggregator and an explicit reason-taxonomy schema would make the output far more usable for automation.This is distinct from #1403 (false-positive on same-file private), #1405 (BCL dominance in hotspots), and closed #105 (implementation of classification). C231 is about exposing the existing taxonomy through a documented schema.
Where
src/CodeIndex/Models/QueryResults.cs:39-44(UnusedSymbolResultfields)src/CodeIndex/Mcp/McpToolHandlers.cs:1444-1447(response serialization)Suggested approach
(1) Document the bucket enum values in
USER_GUIDE.mdand the MCP tool description. (2) Add an aggregator fieldsummary: { by_bucket: {...}, by_confidence: {...} }to the JSON envelope. (3) Add--exclude-tests,--bucket=…filter flags so callers can drill into one classification at a time without re-querying.