By howells
Guides the full software development lifecycle from idea to shipped code: structured ideation, TDD implementation, automated testing, expert code review, mechanical audits, deployment checks, and npm publishing with atomic commits.
Run mechanical verification and comprehensive codebase audit.
Smart commit, push, and npm publish with auto-splitting across domains.
Turn ideas into validated feature specs through collaborative dialogue with expert review.
Risk-proportional planning and implementation with work-kind evidence.
Build and reconcile a vetted backlog of improvement plans.
Whole-implementation specification axis. Compares the complete attributable target with the plan and user request for omissions, partial wiring, stubs, and scope creep.
Whole-implementation standards axis. Reviews maintainability, boundaries, repository conventions, test effectiveness, and work-kind evidence on the complete attributable target.
Use when a test fails unexpectedly during implementation. Investigates root cause systematically, distinguishes between test bugs and implementation bugs, and applies minimal fixes. Prefers event-based solutions over timeout increases. <example> Context: Test fails with timing-related error during implementation. user: "Test 'should complete batch' is failing with timeout" assistant: "I'll use the debugger to investigate the root cause" <commentary> Timing failures need systematic investigation, not timeout increases. Debugger will trace the issue. </commentary> </example> <example> Context: Multiple tests failing after a refactor. user: "3 tests in user-service.test.ts are now failing" assistant: "Let me dispatch the debugger to investigate these failures" <commentary> Concentrated failures in one file suggest a common root cause. Debugger will identify it. </commentary> </example>
Runs and fixes E2E tests (Playwright). Handles flaky tests, timing issues, and selector problems. Iterates until green or reports blockers. Keeps verbose output contained. <example> Context: E2E tests created as part of implementation. user: "Run the e2e tests for checkout flow" assistant: "I'll dispatch e2e-runner to run and fix any issues" <commentary> E2E tests produce verbose output. e2e-runner handles iteration and reports summary. </commentary> </example> <example> Context: E2E tests failing after UI changes. user: "E2E tests are broken after the redesign" assistant: "Let e2e-runner investigate and fix the selector issues" <commentary> UI changes often break selectors. e2e-runner will update them systematically. </commentary> </example>
Writes E2E tests with Playwright. Tests complete user journeys in real browsers — signup flows, checkout processes, authentication. Includes auth setup for Clerk and WorkOS. <example> Context: New feature needs end-to-end coverage. user: "Write E2E tests for the onboarding flow" assistant: "I'll dispatch e2e-test-writer to create Playwright tests for the full journey" <commentary> Multi-page user journey = E2E test. Real browser, real interactions. </commentary> </example>
Create or maintain a project's root CONTEXT.md — goals, product boundary, non-goals, decision principles, and a domain language glossary. Use when starting a new project, clarifying product direction, aligning a codebase for future agent work, defining a north star, pinning down domain terminology, or turning a vague idea into a durable project foundation.
Use when starting any conversation - establishes Arc's skill routing, instruction priority, and bootstrap rules
Comprehensive codebase audit with verification and specialized reviewers. Generates actionable reports. Use when asked to "audit the codebase", "review code quality", "check for issues", "security review", or "performance audit". By default, run the complete audit: mechanical checks first, then specialist reviewers, then a scored report.
Smart commit, push, and npm publish with auto-splitting across domains. Creates atomic commits. Use when asked to "commit", "push changes", "publish", "save my work", or after completing implementation work. Automatically groups changes into logical commits.
Internal plan writer. Produces risk-proportional implementation plans with agreed seams, coherent slices, work-kind evidence, exact paths, and durable lifecycle metadata.
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimnpx claudepluginhub howells/arc --plugin arcBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
The full arc from idea to shipped code.
Arc is a self-contained software development lifecycle for coding agents. It helps move work from early project clarity through ideation, implementation, review, testing, launch readiness, and clean commits.
/arc:* commands.$<skill-name>..cursor-plugin/; install locally or from a team marketplace, then invoke with /<skill-name>.Arc's canonical product definition and operating boundary live in CONTEXT.md. This README is the user-facing guide.
WHY /arc:vision - Define goals, boundary, and domain language in CONTEXT.md
↓
WHAT /arc:ideate - Turn an idea into a validated feature spec
↓
DO /arc:implement - Plan and build with TDD and verification
/arc:testing - Backfill safety-net tests around existing code
/arc:launch - Check go-live and shareability basics
REVIEW /arc:review - Review a plan, spec, or implementation approach
/arc:audit - Verify and audit current codebase health
/arc:refactor - Find structural friction and propose refactors
/arc:improve - Turn findings into a vetted plan backlog, keep it reconciled
TOOLS /arc:commit - Create clean commits, optionally push
/arc:release - Version, changelog, verify, and publish packages
using-arc and detail are supporting skills. They keep startup context small and create implementation plans, but they are not normally invoked directly.
Arc owns the software development cycle:
Arc deliberately does not own every adjacent specialist practice. Brand identity, visual design systems, dependency upgrade campaigns, deep SEO, AI framework guidance, generated documentation, browser QA, and machine/editor automation are better handled by dedicated tools. Arc may notice those concerns during implementation, review, audit, or launch, but its public workflow stays focused on software delivery.
claude plugin marketplace add howells/arc
claude plugins install arc@howells
This installs the full Arc plugin: skills, slash commands, agents, references, disciplines, templates, scripts, and rules.
Native plugin install (Codex CLI 0.117+):
codex plugin marketplace add howells/arc
codex plugin add arc@howells
This installs Arc into Codex's isolated plugin cache — skills, agents, references, disciplines, templates, scripts, and rules — so it never collides with your other skills. Invoke workflows with $<skill-name>; update with codex plugin marketplace upgrade.
Alternatively, a clone-and-symlink installer links Arc skills into ~/.agents/skills (Codex's user skills directory); see .codex/INSTALL.md. Prefer the native plugin above.
Arc is a first-class Cursor plugin (Cursor 2.5+). Local and team installs use the full repo, so skills, commands, agents, references, disciplines, templates, and scripts stay available — the same full-runtime model as Claude Code and Codex. Arc's internal rules/ corpus is not injected as always-on Cursor rules.
# from a checkout
bash .cursor/install.sh
# or clone straight into Cursor's local plugin dir
git clone https://github.com/howells/arc.git ~/.cursor/plugins/local/arc
Reload Cursor (Developer: Reload Window), then invoke /ideate, /implement, /audit, etc. On Teams/Enterprise, import github.com/howells/arc as a team marketplace from the Cursor dashboard. Details: .cursor/INSTALL.md.
npx skills add howells/arc
This copies SKILL.md prompts to supported agents. It is useful for lightweight guidance, but it does not include Arc's bundled agents, references, disciplines, templates, scripts, or rules. Workflows such as audit, review, implement, refactor, and testing work best with the Claude plugin, native Codex plugin, or Cursor plugin.
Use slash commands:
Write complete novels with Claude and Codex. Literary prose, iterative review, and craft-aware feedback.
Development toolkit for Claude Code — plan, implement, ship, review, and assess features with AI-assisted workflows. Progressive zero-config init: auto-configures with sensible defaults on first skill invocation, no upfront ceremony required. Three-tier ceremony model: swift (lightweight), standard (mid-ceremony spec-plan-execute), and thorough (full pipeline) with severity-aware scope routing. Five entry points: arn-planning (scope router, spec, plan), arn-implementing (execute plans, swift, or standard changes), arn-shipping (commit, push, PR), arn-reviewing-pr (PR feedback), arn-assessing (codebase health). Includes arn-code-sketch for UI preview, arn-code-swift for quick implementations, and arn-code-standard for mid-ceremony changes. Includes arn-code-catch-up for retroactive documentation of out-of-pipeline commits. Pipeline preference persistence for streamlined repeat sessions. Batch pipeline: arn-code-batch-planning (multi-feature planning), arn-code-batch-implement (parallel worktree execution), arn-code-batch-merge (conflict-aware merge), arn-code-batch-simplify (cross-feature quality).
Use Claude Code As Is - native plugin leveraging built-in architecture
Standalone tools: codebase audit, test audit, skill evaluation, and usage-window scheduling (queue / usage-guard)
Phase-based AI development framework with 16+ specialized agents, structured phases, and file-based handoffs. Works with greenfield and existing codebases.
Full-cycle project development - brainstorm ideas, create specifications, plan architecture, initialize projects, and execute implementation with integrated workflows from superpowers and spec-kit
A local-first SDLC workflow harness — structured, durable state for coding agents, with convergence gates, agent teams, and full audit trail.