bcmyguest
UserClaude Code skills and agents — install with: npx skills add bcmyguest/personal-skills (or add as a plugin marketplace).
Categories
Indexed Skills (16)
add-ansible-role
Add a new role to the user's local Ansible setup (the playbook that provisions their machine with dev tools — location is read from this skill's config.json). Use whenever the user invokes /add-ansible-role, asks to "add an ansible role for X", "add X to my ansible setup", "manage/install X with ansible", or wants a CLI tool's installation automated the way their other ansible-managed tools are. Also use when updating or fixing an existing role in that playbook, or when the user asks to add shell completion (bash/zsh/fish) for a tool the playbook installs.
senior-frontend-refactor
Refactor frontend code as a senior engineer — improves structure, readability, performance, and test coverage while preserving behavior. Use when the user asks to refactor, clean up, simplify, or restructure existing frontend/React/TypeScript code. For scaffolding a brand-new frontend repo, use react-ts-setup instead.
watchlist
Check upstream GitHub issues, PRs, and releases the user is waiting on, and report what moved since the last check. Use whenever the user invokes /watchlist, asks "any movement on that PR/issue", "check the watchlist", "are we still waiting on X", "has that llama.cpp/ollama/lemonade thing landed yet", or asks to start watching / stop watching an upstream issue, PR, or release. Also use when the user says "remind me to check back on this issue" or "add this to the things we're waiting on".
assign-mr-reviewers
Assign reviewers to the user's active GitLab merge requests based on CODEOWNERS. Use this skill whenever the user wants to assign reviewers, add code owners, or get approvals on their MRs/merge requests. Also trigger when the user mentions "who should review", "need reviewers", "assign owners", or anything about getting their MRs reviewed.
debug-lemonade
Diagnose lemonade-server (lemond) problems on the user's Strix Halo box "d" — server not responding on port 13305, models missing from Open WebUI, a model failing to load, slow inference, or ROCm/Vulkan/GPU questions. Use whenever the user invokes /debug-lemonade or reports anything broken or weird with lemonade, lemond, Open WebUI models, llama.cpp backends, GPU page faults, or "models gone" on this machine, even if they don't name lemonade explicitly (e.g. "open webui shows no models", "the model won't load", "inference hangs").
efficiency-audit
Ruthlessly audit a codebase for its 1-3 most serious EFFICIENCY problems — the hot paths that dominate latency, cost, or resource use — then verify each with a concrete cost analysis (complexity, query count, allocation, growth path) and a fix. Use whenever the user invokes /efficiency-audit or asks to "find the performance bottlenecks", "why is this slow", "audit this for efficiency", "what's burning CPU/memory/money here", "find the N+1 queries", or wants a sharp, high-signal performance review that stops at the top few offenders rather than an exhaustive list. Spins up a small, capped multi-agent workflow of Sonnet agents to trace hot paths and prove cost. For security vulnerabilities use security-audit instead.
pre-commit-setup
Stand up pre-commit in a repository from shipped config fragments — the standard hygiene hooks (whitespace, EOF newline, YAML/JSON/TOML checks, large-file and merge-conflict guards) plus an optional Conventional Commits layer — then install the git hooks and run across the repo. Also the canonical policy for updating hooks safely (autoupdate --freeze to immutable SHAs). Use when adding pre-commit to a project, asked to "set up pre-commit" or "add commit hooks", when updating/refreshing hook pins, or when called from a project-setup skill (rust-setup, uv-setup, react-ts-setup) that layers its language hooks on top.
react-ts-setup
Scaffold a new React + TypeScript frontend repo the right way with Vite — asks the key decisions up front (package manager, Node manager + version, UI framework + theme colours, CI provider, dev Docker), then wires Vite + React Compiler, ESLint (flat) + Prettier, Vitest + Testing Library, husky + lint-staged + pre-commit hooks, .vscode settings, env templates, a dev Dockerfile, CI pipelines, and a README — keeping the Node version in lockstep across every file and leaving lint/typecheck/test/build green. Use this whenever starting a new frontend, React, or TypeScript web project, scaffolding a single-page app, or asked to "set up", "bootstrap", or "initialize" a React/Vite repo. For refactoring existing frontend code, use senior-frontend-refactor instead.
release-setup
Automated releases from Conventional Commits with git-cliff — owns the canonical cliff.toml (commit types → semver bump → release notes) shared by every release flow in this repo, and fully implements the GitLab CI variant (tag vX.Y.Z + GitLab Release with generated notes on every default-branch push). Use when asked to automate releases, tagging, changelogs, or release notes — on GitLab directly; for GitHub the language setup skills (rust-setup, uv-setup) ship the workflow and copy cliff.toml from here.
rust-setup
Scaffold a new Rust repo by copying shipped template files — CI (fmt, clippy -D warnings, locked tests), pre-commit + Conventional Commits, Dependabot, git-cliff auto-release on merge to main with static binaries, crates.io Trusted Publishing (never long-lived tokens), and a tested hello-world mock (CLI bin+lib or pure library). Use only when explicitly asked to scaffold, bootstrap, or set up a new Rust repo, crate, or CLI project — invoke deliberately, e.g. "/rust-setup" or "set up a new rust repo". Not for working in existing Rust projects, and not a trigger for casual Rust questions.
security-audit
Ruthlessly audit a codebase for its 1-3 most serious SECURITY vulnerabilities, verify each one adversarially, and report it with a CVE reference (for known-vulnerable dependencies) or a minimal proof-of-concept (for novel bugs). Use whenever the user invokes /security-audit or asks to "security audit this", "find the worst vulnerabilities here", "is this exploitable", "find the scariest security holes", "what's the biggest security risk in this repo", or wants a sharp, high-signal security review that stops at the top few issues rather than an exhaustive checklist. Spins up a small, capped multi-agent workflow of Sonnet agents to hunt attack surface and prove exploitability. For performance/cost problems use efficiency-audit instead.
uv-develop
Fixes for uv operations an agent otherwise gets wrong — running Python through uv instead of bare python/pip/pytest, upgrading one package without churning the rest, resolving "lockfile out of date" / stale-lock CI failures, and repairing a broken or drifted .venv. Use whenever working in a uv project, i.e. any repo containing a uv.lock (or [tool.uv] in pyproject.toml) — especially before running Python scripts, tests, or tools there; and when upgrading, bumping, or pinning a dependency, when uv reports the lockfile is stale or out of date, or when the venv is broken, drifted, or has untracked packages.
uv-setup
Scaffold a new Python project with uv by copying shipped template files — uv init --package, then overlay templates for ruff + pyrefly (local hook) pre-commit, Conventional Commits, CI entirely through uv, Dependabot (uv + actions), git-cliff auto-release on merge, PyPI Trusted Publishing (pending publisher, never a token), and a multi-stage Dockerfile for services. Use when starting a new Python project, scaffolding a repo, or asked to "set up" or "initialize" a Python project. For day-to-day work in an existing project, use uv-develop instead.
fast-hypothesis
Rapid bug diagnosis — form one best hypothesis and test it with 1-2 commands before exploring further. Use when the user invokes /fast-hypothesis, asks for a quick diagnosis, or wants a bug looked at without long exploratory command chains; for full reproduce-then-verify discipline on a fix, use reproduce-bug instead.
reproduce-bug
Debug any bug using strict reproduce-verify discipline — reproduce the exact failure first, test one hypothesis at a time, and never claim a fix without re-running the original failing case. Use when the user invokes /reproduce-bug, asks to fix a bug properly, or when a fix must be proven against the original failure; for a fast first-pass diagnosis without a fix loop, use fast-hypothesis instead.
writing-with-clarity
Enforces absolute lexical precision, directness, and brevity in agent generation. Use this skill when producing long-form text, structuring explanations, or when an output requires high information density without conversational fluff.
Bio shown is the top-scored skill's repo description as a fallback — real GitHub bios land in a future update.