Architecture Risk Assessor

This skill appears coherent: it analyzes architecture descriptions and repository artifacts and outputs risk matrices and mitigation notes. Before installing or enabling it, consider these practical points: - The skill will try to read project files (docker-compose, k8s manifests, package files, READMEs) if available — grant only read-only access and limit scope to directories you want analyzed to avoid exposing secrets or unrelated configs. - It requests no credentials and does not install software, so its technical footprint is small; however, review any assessment output before sharing externally because it may summarize sensitive architecture details. - Note the skill's 'Unproven Technology' rule: it prescribes assigning maximum risk (9) to technologies not used in production. This is a strict, deliberate policy that can bias results toward high-risk for unfamiliar tech — you may want to override or contextualize that rule when appropriate. - If you do not want the agent to scan your repo, provide an explicit architecture description instead. Overall this skill is internally consistent with its stated purpose; treat it as a read-only analysis assistant and restrict file-scan scope if you have sensitive project data.