Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →ellmos-ai/bugsweep
Systematic bug sweep with a codebase-scaled target value, doubling escalation, area tracking, and final verification.
Passed 12 of 17 test cases with the skill, and 5 of 17 without it.
and 14 more in the full report
13 fixed3 passed either way3 still failing3 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 10 test cases the skill fixed. Same model, same prompt, with and without the skill.
We want to check `src/api/auth.py` in run 1, and then expand to `src/api/auth.py` and `src/api/user.py` in run 2. Specify the exact pre-search area declaration format for run 1.
| Checked | Without skill | With skill |
|---|---|---|
| Formats the area declaration using the pattern '"{Name}" (code)' or '"{Name}" (code|aufgabe)'. | Missed | Passed |
The instructions your model reads when the skill loads.
Iterative Bug-Suche mit konvergierendem Abbruchkriterium. Skaliert mit der Codebasis, eskaliert bei Verdacht auf oberflächliche Suche, und verhindert Wiederholung durch Bereichs-Tracking.
LOC = produktive Quellzeilen (src/, lib/ — ohne Tests, Configs, Docs, generated)
x = max(1, ceil(LOC / 1500))
Grundrate = x * 3
| LOC | x | Grundrate |
|---|---|---|
| ~1500 | 1 | 3 |
| ~3000 | 2 | 6 |
| ~4500 | 3 | 9 |
| ~10000 | 7 | 21 |
Melde dem User: "Codebasis: {LOC} LOC → Grundrate = {Grundrate} saubere Suchläufe."
zähler = 0
ziel = Grundrate
je_bug_gefunden = False
geprüft = [] # (bereich_name, typ: code|aufgabe)
LOOP:
bereich = wähle_neuen_bereich() # Siehe Bereichs-Regeln
geprüft.append(bereich)
Führe gründliche Bug-Suche durch
IF Bug gefunden:
je_bug_gefunden = True
Fix nach bugfix-protocol (Phase 4+5)
test cases passed, vs 5 / 17 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.