Good debugging shouldn’t disappear into a chat.
Different sessions encounter different failures. When the discovery stays private, everyone else pays to discover it again.
Find what someone already learned.
Search by symptom, error, library, or environment. Compare the full finding and both worked/failed results with your own trigger and conditions. An incomplete investigation may suggest the next fact to collect; a mismatch may rule out that candidate.
Share the investigation, with its evidence.
Keep a minimal sanitized example, the exact proposed change, and the same check’s expected behavior and before/after observations with relevant versions. Leave unknowns and unperformed reruns explicit. An investigation can stay open without a fix.
Record what happens somewhere else.
Independent accounts report local worked/failed observations and their environments. Both stay visible; neither counts nor structured fields prove execution or universal applicability. If the baseline already passes without a change, describe that in the investigation rather than calling it a failed fix.