Custom Claude Code skills for my development workflow. Built for how I work; fork and adapt for yours.
Each skill is a directory with a SKILL.md that Claude Code discovers automatically. Invoke with /skill-name inside any Claude Code session.
Stress-test a plan or design through relentless Socratic questioning. Builds a risk tree, walks it depth-first, and won't move on until every branch is resolved. Dispatches research agents when you say "I don't know" instead of letting you off the hook.
/grill-me when you want your design torn apart before production does it for you.
Systematically address every review comment on a GitHub PR. Categorizes comments by priority, fixes code, runs tests, resolves threads via GraphQL, and updates the PR description. Nine-phase workflow that handles the tedious loop of review-fix-verify-push.
/fix-pr-reviews when a PR has review comments piling up.
Extract structured context from a planning session and generate a handoff document for a fresh build session. Captures decisions, negative knowledge ("Do Not Retry"), constraints, and next steps. Uses a JSONL preprocessor and Claude-as-extractor architecture. Auto-saves session transcripts for eval corpus building.
Includes a blind eval harness (/eval-extract) that dispatches isolated subagents to grade extraction quality across 8 dimensions.
/handoff when a planning session is complete and you want to start building in a clean context window.
Full codebase planning pipeline using Claude Code Agent Teams. Spawns 4-9 analyst agents to scan a repo in parallel, interviews you using their findings, validates scope, and outputs a prioritized ROADMAP.md plus GitHub Issues with dependency links. One session, ~20-30 minutes.
/scope-repo when you want a structured roadmap for a repo you're about to invest serious time in.
Requires Agent Teams enabled (CLAUDE_CODE_EXPERIMENTAL_AGENT_TEAMS=1 in settings.json). Also ships a standalone CLI wrapper (scope-repo/scopework) for running outside a session.
Symlink individual skills into your Claude Code skills directory:
git clone https://github.com/brianruggieri/skills.git ~/git/skills
# install one skill
ln -s ~/git/skills/grill-me ~/.claude/skills/grill-me
# install all
for skill in ~/git/skills/*/; do
[ -f "$skill/SKILL.md" ] && ln -s "$skill" ~/.claude/skills/$(basename "$skill")
doneOr copy a skill directory into any project's .claude/skills/ to share it with collaborators.
grill-me/
└── SKILL.md
fix-pr-reviews/
└── SKILL.md
handoff/
├── SKILL.md
├── prompts/ # extraction + convention detection prompts
├── eval/ # blind eval harness (SKILL.md + grading rubric)
├── scripts/ # JSONL preprocessor
└── tests/ # fixtures for eval
scope-repo/
├── SKILL.md
├── scopework # standalone CLI wrapper
└── install.sh
Each skill is self-contained in its directory. The SKILL.md has YAML frontmatter (name, description, allowed-tools) and a markdown body with the full protocol Claude follows when the skill is invoked.
MIT