buildproven

ralph

Autonomous backlog execution with reflection, evidence, worktree isolation, and /bs:quality --merge

recover-quality

Comprehensive audit + fix for quality regression across all repos; integrates learnings into CLAUDE.md

Runtime agent observability — pull recent Sentry errors, cluster by signature, file Linear tickets with stack traces, optionally open worktree + PR with proposed fix. Plus synthetic user-journey cron orchestration.

verify-claim

Enforce research rigor — extract factual claims from drafts, verify each against primary sources, mark unverifiable as [unverified] before output. Use before writing any research, strategy, or competitive-analysis memo.

agent-browser

Browser automation CLI for AI agents. Use when the user needs to interact with websites, including navigating pages, filling forms, clicking buttons, taking screenshots, extracting data, testing web apps, or automating any browser task. Triggers include requests to "open a website", "fill out a form", "click a button", "take a screenshot", "scrape data from a page", "test this web app", "login to a site", "automate browser actions", or any task requiring programmatic web interaction.

API & Backend Listed

api-conventions

Auto-invoke skill for consistent API design conventions. Activates when writing API routes, HTTP handlers, response objects, or REST endpoints. Enforces RESTful naming, consistent response shapes, proper status codes, Zod validation at boundaries, pagination, and rate limiting patterns.

deps

Dependency health — outdated packages, security audit, smart upgrades

error-handling

Auto-invoke skill for consistent error handling patterns. Activates when writing try/catch blocks, API error responses, form validation, async operations, or error boundaries. Enforces consistent patterns and flags anti-patterns like empty catches and silent failures.

Web & Frontend Listed

frontend-design

Create distinctive production-grade product UI, marketing pages, and redesigns. Use when building or styling frontend components, pages, or applications; routes by interface type and iterates rendered output to avoid generic AI aesthetics.

merge-train

Parallel cross-repo quality+merge sweep. One Task agent per repo in an isolated worktree runs /bs:quality --merge, auto-merges if green, post-merge cleanup, consolidated summary. Honors the never-skip-pre-existing-broken-CI rule.

new

Create new project with QA Architect quality automation (framework-agnostic)

patterns

Search CLAUDE.md for patterns and best practices across all projects

prd

Carson-pattern PRD discipline — 3-5 clarifying questions before draft, junior-dev-targeted PRD, dot-notation task list (0.0 = create feature branch), machine-verifiable acceptance criteria, mandatory pause gates between parent tasks and subtask expansion. The forcing function for any work that takes more than one day.

quality

Autonomous, revision-bound quality loop: deterministic gates, independent review, CI, and optional merge.

recover

Crash and state recovery skill. Auto-invokes when Claude Code crashes, loses context, encounters broken state, won't start, hangs, freezes, hits stale locks, or has incomplete git operations. Guides through diagnosis, recovery, and context restoration.

review-content

Collaborative artifact review skill for products, PDFs, landing pages, email sequences, decks, and long-form docs. Use when you want a serious pre-ship critique focused on SOTA quality, differentiation, clarity, conversion strength, and competitive sharpness rather than code quality.

scrub

Prepare a project for public release - open source, free giveaway, or commercial sale. Auto-invokes on natural language like "get this ready for open source", "scrub for release", "clean before publishing", "prep for giveaway", or "prepare commercial release". Removes secrets, dev infrastructure, internal references; generates mode-appropriate LICENSE and docs.

sota

SOTA system assessment skill. Auto-invokes when user asks "how's my setup?", "rate my system", "am I SOTA?", "score my config", "audit my Claude setup", or similar assessment questions. Provides consistent, repeatable scoring against defined benchmarks.

test-strategy

Auto-invoke skill for behavioral test strategy. Activates when changing functions, components, APIs, workflows, or shared modules. Chooses the highest useful public interface and seam, requires red-capable tests with independent expected values, and avoids implementation-coupled file-per-source coverage.

workflow

Daily development workflow reference. Quick-start guides for solo dev, team collaboration, production launches, and emergency hotfixes. Points to detailed steps in supporting files.

codebase-design

Design and refactor deep modules with small interfaces, clean seams, high caller leverage, and tests through public behavior. Use for architecture, refactoring, interface design, testability problems, pass-through abstractions, or scattered change.

codex

Codex health: ChatGPT auth, native skills, canonical AGENTS, and MCP profile

dev

Start development work (features, bugs, refactoring, experiments)

diagnosing-bugs

Disciplined diagnosis loop for hard bugs, regressions, flaky behavior, and performance failures. Use before implementing a fix: create a tight red-capable command, minimize the reproduction, test ranked falsifiable hypotheses, then lock the fix with behavioral evidence.

hotfix

Fast-track production emergency fixes (5-10 min minimal quality)

legal

Legal document review and advisory for software founders and indie SaaS builders. Reviews privacy policies, ToS, EULAs, SaaS agreements, API terms, open source licenses, NDAs, and contractor agreements. Flags software-specific risks (IP assignment, data ownership, warranty disclaimers, copyleft traps, AI output rights). Provides plain-English explanations with actionable recommendations. Not a substitute for a licensed attorney.

review

Collaborative artifact review for products, PDFs, landing pages, emails, and docs via the shared ensemble runner

seo

SEO keyword research, SERP analysis, backlink checks, and competitor intelligence using DataForSEO API.

strategy

Multi-model strategy synthesis & advisory panel via shared ensemble runner

ui-reviewer

Strict UI/UX reviewer for screenshots or live product and marketing interfaces. Scores product specificity, generic AI-pattern risk, hierarchy, typography, color, responsive states, interaction quality, and accessibility; provides concrete fixes and a 99/100 ship threshold.

steward

Discover recently active repositories, audit convergence, and optionally repair findings through a provider-neutral PR workflow