I dogfooded my own code intelligence tool and found 4 bugs I'd shipped

Building Sverklo (launching tomorrow on PH), I ran a structured dogfood protocol — used the tool on its own codebase to find real bugs before users did.                                                                                  

  Found 4 integration-level bugs that unit tests missed:                                                                                                                                                                                                                                                                                                                                                                                  

1. Impact analysis silently dropped repeat call sites — the worst possible failure for a refactor-safety

2. Reference search returned 48 substring matches, drowning the 5 real

3. Lookup returned "No results" on valid queries instead of explaining why    

4. Parser off-by-one skipped every function after the first in multi-function                                                                                                                                                                                                                                             

5. All fixed, regression-tested, and documented in a full unedited session log

Curious — has anyone else here done a structured dogfood protocol on their own tool before shipping? I found it caught things no synthetic test could, and I'm thinking of making it a standard part of every release.                    

Would love feedback from anyone who tries it — especially if something breaks. I triage issues within hours.

Submit it — this builds anticipation before the launch goes live.