The Agent Skills Directory

[SAFE]: The skill defines a set of procedures for evaluating and scoring software architecture dimensions.
It emphasizes "reproducible" and "defensible" scoring.
It enforces a "confidence cap" rule to prevent high scores based on thin evidence.
[SAFE]: The skill references standard development and static analysis tools.
Mentions tools like scip, codegraph, lizard, and jscpd as legitimate sources for gathering architectural evidence.
[SAFE]: No network access, credential exposure, or prompt injection patterns were identified.

architecture-scorecard