What is this?
Entry
What does it cover that the current corpus does not?
For Vulture, src/todo/domain/status.py naturally exercises jendrikseipp/vulture#312: passing a bare enum class to list() marks its members as used. The historical comparison drops 3 findings (BACKLOG, TODO, and IN_PROGRESS), from 7 to 4.
The module-level target is intentional. A package-wide Vulture scan sees references to the enum members elsewhere and masks this behavior.
For Skylos, scanning the broader src/todo package with danger, secrets, quality, and ai-defects finds four SKY-D211 SQL-injection patterns. This rule ID is absent from the current all-modes corpus. These are direct stdlib SQLite shapes, complementary to the SQLAlchemy shapes in the proposed Meltano SDK entry.
Size and maintenance
GitHub reports about 800 KB. The repository was created in April 2026 and was active at the proposed pin on 2026-08-18. It is small and current, although its maintenance history is necessarily shorter than that of the other proposed entries.
Measured two-sided detector cost (--jobs 1, base plus head):
- Vulture module target: about 0.07 seconds; proposed corpus cost: 1 second.
- Skylos
src/todo, dead code plus all four optional analyses: about 7.13 seconds and 264 findings per side; proposed corpus cost: 10 seconds.
Anything else
Suggested per-tool settings:
tools:
vulture:
targets:
- src/todo/domain/status.py
cost: 1.0
skylos:
analyses:
- danger
- secrets
- quality
- ai-defects
targets:
- src/todo
cost: 10.0The exact historical #312 replay required Python 3.12 plus a compatibility wrapper because those Vulture revisions predate current PEP 621 project metadata handling. They produced parseable findings but a nonzero detector exit, so the 3-finding differential is supporting evidence rather than a clean managed liveness_primer result.
What is this?
Entry
What does it cover that the current corpus does not?
For Vulture,
src/todo/domain/status.pynaturally exercises jendrikseipp/vulture#312: passing a bare enum class tolist()marks its members as used. The historical comparison drops 3 findings (BACKLOG,TODO, andIN_PROGRESS), from 7 to 4.The module-level target is intentional. A package-wide Vulture scan sees references to the enum members elsewhere and masks this behavior.
For Skylos, scanning the broader
src/todopackage withdanger,secrets,quality, andai-defectsfinds fourSKY-D211SQL-injection patterns. This rule ID is absent from the current all-modes corpus. These are direct stdlib SQLite shapes, complementary to the SQLAlchemy shapes in the proposed Meltano SDK entry.Size and maintenance
GitHub reports about 800 KB. The repository was created in April 2026 and was active at the proposed pin on 2026-08-18. It is small and current, although its maintenance history is necessarily shorter than that of the other proposed entries.
Measured two-sided detector cost (
--jobs 1, base plus head):src/todo, dead code plus all four optional analyses: about 7.13 seconds and 264 findings per side; proposed corpus cost: 10 seconds.Anything else
Suggested per-tool settings:
The exact historical #312 replay required Python 3.12 plus a compatibility wrapper because those Vulture revisions predate current PEP 621 project metadata handling. They produced parseable findings but a nonzero detector exit, so the 3-finding differential is supporting evidence rather than a clean managed liveness_primer result.