
bntvllnt/agent-skills
8 skills341 installs120 starsGitHub
Install
npx skills add https://github.com/bntvllnt/agent-skillsSkills in this repo
1WorkflowWorkflow is a Claude skill implementing a high-velocity solo development loop from idea to production in a day. It exposes 10 commands (plan, spike, ship, fix, review, spec-review, focus, done, drop, workflow) and auto-detects intent from context. It is spec-first and TDD-enforced with quality gates for lint, typecheck, build, test, E2E, and coverage. Developers use it to structure feature and bug-fix work with specs and regression tests while keeping deployment under human control.71installs2Analyzeanalyze is a skill that runs multi-perspective analysis of any topic, file, idea, or decision. A developer invokes it to extract key points, surface gaps and risks, and generate an actionable improvement roadmap. It has quick, standard, and deep modes that dispatch 0-10 parallel perspective sub-agents, and for TypeScript code paths it can augment its reasoning with the codebase-intelligence CLI. It challenges assumptions with a first-principles brief.61installs3GitGit is a Claude skill that runs a repeatable git workflow biased toward safety. It routes user intent (status, commit, branch, worktree, tag, PR, review, merge/rebase) to the smallest matching reference procedure. Developers use it to stage and commit safely, manage worktrees, and create or review pull requests via the gh CLI. It enforces confirmation before any state- or history-changing operation.58installs4GithubGitHub is a Claude skill that operates GitHub through the gh CLI. It routes intents to references for repos, issues, pull requests, Actions workflows, CI monitoring, releases, secrets and variables, projects, gists, search, and raw gh api access. Developers use it to triage issues, create and review PRs, watch CI, and publish releases. Read-only commands run freely while state-changing operations require confirmation.56installs5Skill BuilderSkill Builder is a Claude skill for authoring Agent Skills for any compatible agent product. It routes each request to the smallest operation - create, update, add, delete, rename, or validate - and runs a first-principles check before writing anything. Developers use it to build correct, spec-conformant SKILL.md files with clear routers, safety rules, and progressive disclosure. It returns validation evidence rather than relying on vibes.42installs6Tmuxtmux is a Claude skill for managing the tmux terminal multiplexer. It routes intents to references for session, window, and pane management, layouts, copy mode, configuration, keybindings, and scripting or automation via send-keys. Developers use it to create and attach sessions, split and resize panes, capture pane output, and edit tmux.conf. It confirms before killing sessions, windows, or panes and warns about running processes.24installs7Convexconvex is a skill for building and operating Convex backends, including queries, mutations, actions, HTTP actions, schemas, auth, scheduling, file storage, components, migrations, performance, and testing. A developer invokes it when working in a repo with a convex/ directory. It delegates to the official get-convex/agent-skills collection when available and enforces blocking rules to verify docs and runtime behavior before shipping. It also covers multi-environment and parallel-worktree dev setups.22installs8Oss ReadinessOSS Readiness is a Claude skill that gates a repository for public open-source release. It audits against a 23-item checklist, scaffolds missing public docs and templates, generates llms.txt and llms-full.txt, validates CI, and syncs version references. Developers use it before going public to catch missing licenses, contributing guides, security policies and stale version numbers. It stays harness- and maintainer-agnostic and uses placeholders instead of hardcoded links.7installs