Agent Skills by sunbluesome
Showing 24 vetted skills indexed across 1 GitHub repositories.
release-note
Generate release notes for model versions from project charter and source code.
git-commit
Generate a squash-commit message summarizing branch changes since main.
github-pr
Generate a pull request with an English title and Japanese description from branch changes.
Skill Development
Automate creation and validation of Claude Code plugin skills with structured frontmatter.
weekly-report
Summarize weekly project activity into communication briefs and IMRaD reports.
interface-spec
Generate structured ML interface specifications with schemas and CLI arguments for SageMaker Processing Jobs.
cli-entrypoint
Build environment-aware CLI entrypoints for data-processing pipelines with Click decorators.
code-review
Audit codebases for design principles, architecture, and type-safety compliance.
project-charter
Structure project charters from stakeholder interviews into formal drafts.
robustness-analysis
Identify defense gaps across L1-L4 layers in data processing pipelines.
bdd-testing
Generate Gherkin features and pytest-bdd step definitions for BDD testing workflows.
ml-system-design
Design production ML systems to prevent training-serving skew and data leakage.
processing-flow-doc
Document complex processing flows with C4 model diagrams and ADRs.
dev-cycle
Coordinates end-to-end development workflows across Phase 0-7 with approvals and quality gates.
jupyter-experiment
Organize Jupyter notebooks into IMRaD-structured single-hypothesis experiments.
readme-generator
Generate a README.md by inspecting charter docs, source code, and dependencies.
bdd-test-review
Audit BDD feature files and step definitions for coverage gaps.
coding
Automate Python component implementation with DTO-based contracts.
team-code-review
Coordinate team-based code reviews with architect and category reviewers.
team-code-refactor
Coordinate architect-directed code refactoring within an existing review team using pyright validation.
experiment-doc
Generate IMRaD-formatted experiment documentation from research notes and results.
test-coverage
Identify and document test coverage gaps in AI-generated test suites.
project-knowledge
Organize and retrieve project domain knowledge from _docs/knowledge.
review-feedback
Collect code review decisions and generate consolidated action plans.