You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
-**Logic:** Extracted normalization, host matching, and exclusion logic from `ResultConsumer`.
23
+
-**Benefit:** Improved testability (unit tests added) and separation of concerns.
24
+
25
+
### 4. MCP Error Handling
26
+
-**Fix:**`qurio_search` (and others) now return standard JSON-RPC Error objects (`code`, `message`) instead of embedded success results with `IsError: true`.
Plan updated with clean formatting and explicit target descriptions.
2
-
`qurio_list_pages` uses `source_id`.
3
-
`qurio_read_page` uses `url`.
4
-
Full argument guide included for `qurio_search`.
5
-
6
-
## Ingestion Worker
7
-
-`handle_file_task` now returns `list[dict]` (standardized with `handle_web_task`) to simplify `main.py` logic.
8
-
- Metadata (Title, Author, PageCount) is extracted via Docling v2 and passed to backend.
1
+
-**Logging:** Implemented `apps/backend/internal/logger` package with `ContextHandler` to automatically propagate `correlation_id` from context to JSON logs.
2
+
-**Worker Logic:** Link discovery logic extracted to pure function `DiscoverLinks` in `apps/backend/internal/worker/link_discovery.go` to separate I/O from business logic.
3
+
-**MCP Errors:**`qurio_search` now returns standard JSON-RPC errors (code -32603) for internal failures instead of embedded text errors.
0 commit comments