Produces read-only, evidence-backed codebase analysis with ranked feature-sized candidates.
npx skills add https://github.com/pnewsam/skills --skill analyze-pnewsam
Please help me install this Agent Skill. Skill: analyze Source: https://github.com/pnewsam/skills/tree/main/archive/analysis-validation-evicted/analyze Command: npx skills add https://github.com/pnewsam/skills --skill analyze-pnewsam
High-intent search queries and answers about installing and using this skill.
Run a read-only analysis that gathers dependency advisories, SAST findings, and exposure signals, then verifies each finding is real and reachable. The output is a ranked set of feature-sized candidates with cited evidence, ready for a planning step.
Feed scanner output into an analysis pass that discards noise, confirms reachability, and groups findings into bounded candidates ranked by impact versus effort. Each candidate includes the measured evidence and rough size so planning can pick one up directly.
Yes, design-system convergence is a supported dimension. The analysis scans token, primitive, and component-family usage to inventory pattern drift, duplication, and legacy components, then ranks migration candidates by leverage.
No, the analysis is strictly read-only. It never edits code, dependencies, styles, or configuration, and never commits, pushes, or changes external systems, so it is safe to run at any point before planning.
Skip it when you need a formal threat model with assets, actors, and trust boundaries, or a current-state architecture document, since those are different output framings produced directly. It also does not provide legal or regulatory interpretation such as GDPR applicability.
superagent-aiEvaluate security advisories against repository checkouts and assign severity.
microsoftCompare RPI planning and implementation evidence and record severity-graded review findings.
platyplusValidate row counts, taxonomy, and geo-spatial constraints before and after bulk record insertion.
openaiGenerate repository-grounded Markdown threat models with trust boundaries and abuse paths.
