Adam Krysztopa
Community@AdamKrysztopa
Adam Krysztopa publishes architecture-decision skills covering threat modeling, testing strategy, system topology, object-level design patterns, and LLM-agent system design.
Agent Skills by Adam Krysztopa
Showing 5 vetted skills indexed across 1 GitHub repositories.
AdamKrysztopathreat-model
Derives security controls from assets, actors, and trust boundaries through a structured threat interview.
AdamKrysztopatest-patterns
Designs risk-led testing strategies and reviews existing test suites for rebalancing.
AdamKrysztopadecide-architecture
Select or review software architecture through a branching interview and pattern catalog.
AdamKrysztopadesign-patterns
Recommends GoF and Python-idiomatic design patterns for new code or reviews existing code for pattern smells.
AdamKrysztopaagentic-patterns
Designs or audits LLM-agent systems using a layered pattern catalog and autonomy gate.
Frequently Asked Questions About Adam Krysztopa
FAQPage SchemaWhat tasks can I accomplish with Adam Krysztopa's skills?▼
Five decision-support skills: threat-model for architecture security reviews, test-patterns for testing strategy and suite rebalancing, decide-architecture for system topology choices, design-patterns for GoF and Pythonic class design, and agentic-patterns for LLM-agent system design and review.
Who should use these skills?▼
Software architects, senior engineers, and technical leads making or reviewing design decisions. Each skill branches automatically: greenfield work gets a structured selection interview, while existing code gets a review against a pattern catalog with targeted, highest-leverage moves.
How do the skills handle greenfield versus existing code?▼
Greenfield triggers a short interview producing a recommendation, ADR, or evidence portfolio. Existing code triggers a catalog-based review returning one highest-leverage move. Threat-model binds scanner claims to scanners the repository already runs and states plainly when no binding exists.
What does the threat-model skill explicitly not do?▼
It never scans: no CVE enumeration, dependency grading, secret hunting, or scanner simulation. It produces no maturity score, percentage, or checklist. It also excludes agent control-flow design (agentic-patterns), topology selection (decide-architecture), and test selection (test-patterns).
What design philosophy guides the agentic-patterns skill?▼
It pushes toward the least autonomy that works, using a layered interview covering autonomy, reasoning loop, topology, memory, reliability, governance, input boundary, and integration. Existing agent code is reviewed against a nine-point checklist of over-building, under-building, and stale-scaffold defects.