v0.8.0
What's Changed
This release introduces two significant new skills and several bug fixes to strengthen the agent framework.
dotnet-benchmark brings structured performance testing to .NET development. The skill provides discovery-focused benchmarking with deterministic candidate selection, representative workload definition, and automated BenchmarkDotNet experiment generation. It includes comprehensive validation tooling, style enforcement, and coverage requirements to ensure benchmark quality.
agent-smith establishes software-craftsmanship standards across engineering tasks including design, architecture, implementation, refactoring, testing, security, CI/CD, and governance. The skill grounds recommendations in repository conventions and validates changes before claiming completion. See agent-smith documentation for details on its capabilities and integration patterns.
Additional improvements standardize PowerShell execution to pwsh 7+ across the repository, fix release scope boundary handling in git-keep-a-changelog, and add deterministic subject validation to git-visual-commits for consistent commit messaging.
Note
agent-smith is in preview. This skill is under active development as we refine the software-craftsmanship framework. Early feedback on design patterns, compliance rules, and recommendation accuracy is welcome. The skill's behavior and output may change based on usage patterns and community input.
Sources:
- V0.8.0/dotnet benchmark by @gimlichael and @aicia-bot in #31
- V0.8.0/agent smith by @gimlichael and @aicia-bot in #32
Full Changelog: v0.7.5...v0.8.0