howells
- 58 skills
- 0 followers
- 8 hours ago last updated
- ▌ Surface · howells bundleMake software legible to agents and guide production agent-system design. Use when asked for Agent Surface guidance, agent-readable software, agent readiness, AGENTS.md/llms.txt/MCP/OpenAPI/CLI/tool surfaces, agent protocols or tooling, audits or scorecards, transformation plans, or scaffolding agents/tools/workflows/memory/model routing/browser/sandbox capabilities across APIs, CLIs, docs, auth, errors, tests, retrieval, and multi-agent systems.
- ▌ Audit · howellsComprehensive 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.
- ▌ Commit · howells bundleSmart 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.
- ▌ Detail · howellsInternal plan writer. Produces risk-proportional implementation plans with agreed seams, coherent slices, work-kind evidence, exact paths, and durable lifecycle metadata.
- ▌ Ideate · howellsTurn ideas into validated feature specs through collaborative dialogue with built-in expert review. Use when asked to "design a feature", "plan an approach", "think through implementation", or when starting new work that needs product and technical shaping before coding.
- ▌ Launch · howells bundleGo-live and shareability checklist covering the basics needed to make a project visitable, shareable, and ready for a first real audience. Use when asked to "launch", "go live", "make this shareable", "get this ready to show people", or prepare a project for a public URL.
- ▌ Vision · howellsCreate 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.
- ▌ Improve · howellsTurn findings into a vetted, prioritized backlog of executable implementation plans, and keep that backlog alive across sessions. Use when asked to "build an improvement backlog", "plan the audit findings", "what should we improve", "turn these findings into plans", "reconcile the plans", or "what's left in the backlog". Also surfaces grounded direction findings — evidence-based feature candidates the codebase itself suggests.
- ▌ Release · howellsVersioned releases of npm packages: semver bumps, changelogs, and multi-package releases for single-package repos and pnpm workspaces with changesets. Use when asked to "cut a release", "bump the version", "release the packages", "publish a new version", "run changesets", or "ship a versioned package".
- ▌ Testing · howellsCharacterization testing and safety-net backfill for existing code. Use when legacy, under-tested, or risky code needs tests before a refactor, bug fix, or behavior change. Captures current behavior through public interfaces, identifies coverage gaps, and adds focused unit, integration, or E2E tests without replacing TDD implementation workflows.
- ▌ Refactor · howellsDiscover architectural friction and propose structural refactors with competing interface options. Focuses on deepening shallow modules, extracting grouped concerns into packages/modules, breaking up god files, reducing duplication, and improving testability. Use when asked to "improve the architecture", "find refactoring opportunities", "deepen modules", "consolidate coupling", "break up god components", "extract this into a package", "make this more testable", or "find architectural friction".
- ▌ Implement · howellsRisk-proportional implementation workflow with coherent slice ownership, work-kind evidence, durable resumption, whole-implementation review, and one fresh closeout gate. Use when asked to implement, build, execute a plan, or continue from an Arc feature spec.
- ▌ Using Arc · howellsUse when starting any conversation - establishes Arc's skill routing, instruction priority, and bootstrap rules
- ▌
- ▌ Edit · howellsRun line-level editing across chapters. Catches spelling, grammar, awkward phrasing, and word echoes.
- ▌
- ▌ Cover · howellsGenerate book cover art prompts for image generation models. Reads project content and produces optimized prompts for Kindle-dimension covers.
- ▌ Notes · howells bundleCollect and process inline editing notes from chapters. Finds %%NOTE:%%, %%EDIT:%%, %%CHECK:%%, and %%CUT:%% markers.
- ▌ Naming · howellsGenerate and validate book title options. Produces candidates using proven naming strategies and checks availability against existing books.
- ▌ Review · howellsReview chapters for coherence, cheesiness, consistency, pace, tone, and character. Suggests specific rewrites.
- ▌
- ▌ Fiction · howellsThis skill should be used when the user asks to "write a chapter", "write prose", "continue the story", "develop a character", "review my chapter", "critique my manuscript", "write a synopsis", "summarize my story", "plan my novel", "outline my book", "check for consistency", or mentions fiction writing, novels, short stories, scenes, or narrative craft.
- ▌
- ▌ Language · howellsVerify non-English phrases in your manuscript. Checks grammar, period accuracy, dialect, and translations.
- ▌
- ▌ Reconcile · howells bundleAudit a project against current plugin conventions and offer to update structure.
- ▌ Illustrate · howellsGenerate consistent chapter illustrations using motif series. Creates a visual style from the cover, builds character references, then generates one illustration per chapter with consistent styling.
- ▌ References · howells bundleInternal fiction craft reference library for the Fiction plugin. Do not invoke directly; use fiction, review, edit, critique, plan, outline, character, or another task skill that loads the relevant craft references.
- ▌ Gog · howells bundleUse gogcli for configured Google accounts instead of Google connectors. Not for web research (`web-research`).
- ▌ Next · howells bundleClaim and deliver the next item with Plimsoll; name the thread for it. Not audits (`simplify`) or recovery (`muster`).
- ▌ What · howells bundleRe-explain only the previous reply in three plain lines: what happened, current state, next step. No new facts. Not task status (`memento`) or session recovery (`muster`).
- ▌ Xero · howells bundleQuery Xero accounting data through a configured integration. Not for Starling (`starling`).
- ▌ Nomen · howells bundleGenerate, critique, and validate names for products, projects, packages, CLIs, apps, brands, or features, with domain, package, GitHub, App Store and web conflict checks. Use for naming and renaming decisions. Not for interface labels (`signage`), body prose (`deslop`), or legal clearance.
- ▌ Deslop · howells bundleRewrite prose that sounds synthetic, inflated, or assistant-like. Use for AI-writing tells, vague attribution, suspicious citations, chatbot artifacts, or copy that should sound grounded and human. Applies to prose; for interface labels use `signage`; for machine-written code tells use `unslop`.
- ▌ Linear · howells bundleUse configured Linear accounts via GraphQL; choose the account that owns the target. Not for transcripts (`muster`).
- ▌ Muster · howells bundleRebuild context across concurrent or older tasks from transcripts, git, peer sessions and the tracker. Use for catch-ups and in-flight inventories. Not for one current task (`memento`), rewriting the last reply (`what`), or repository cleanup (`salvage`).
- ▌ Unslop · howells bundleRemove machine-written code tells from a diff without changing behavior: narrating comments, one-caller wrappers, impossible guards, unused options, decorative logs, and leftover scaffolding. Use before commit or review. For behavior-changing fallback cleanup use `fail-fast`; for prose use `deslop`.
- ▌ Blender · howells bundleInspect, measure and render Blender scenes through MCP or headless CLI. Use for Blender scene, render-pipeline or MCP work. Not for 2D design files (`paste-up`).
- ▌ Foreman · howells bundleRun an explicitly requested delegation mode for substantial changes: the main agent decides and inspects while subagents write code. Use only when the user asks for Foreman or delegated execution. Delegation alone never requires new tests. Not for ordinary implementation, tiny fixes, or docs-only work; `plimsoll` governs process weight.
- ▌ Memento · howells bundleReport what this task is doing now from live state: original ask, completed work, remaining work, branch, uncommitted files, and drift. Use when returning to a long session or checking whether an agent is still on track. For a roll-call across tasks use `muster`; for rewriting only the last reply use `what`.
- ▌ Salvage · howells bundleRescue work that exists in only one place - detached HEADs, worktrees, unpushed branches, or stashes - then remove only what is proven merged and pushed. Use for repository cleanup where ambiguous work must be preserved and each surviving thread named. For status reconstruction without cleanup use `muster`.
- ▌ Signage · howells bundleRewrite interface strings into the words the audience already uses: labels, headings, buttons, status lines, empty states and generated text. Use when reviewing UI copy for comprehension. Not for body prose (`deslop`), product names (`nomen`), layout or visual styling (`chiaroscuro`), or the type ramp (`typecase`).
- ▌ Paste Up · howells bundleBuild a Paper mockup from a written specification, or audit and repair an existing Paper file: tokens, fonts, artboard layout, crops, labels. Use only for work in app.paper.design. Not for coded UI implementation (`chiaroscuro`), browser QA (`fieldtest`), or design critique with no Paper file involved.
- ▌ Plimsoll · howells bundleCut process weight when gate ladders, CI/build watch loops, remote Vercel builds or re-planning displace shipping. Keeps Vercel builds on the user's machine. Use near a deadline or after nothing user-visible has landed. Not `simplify`.
- ▌ Simplify · howells bundleReview a diff or whole codebase for structural simplification and relevant health risks, with source-confirmed findings and optional scores. Not routine cleanup (`unslop`) or browser QA (`fieldtest`).
- ▌ Starling · howells bundleQuery configured Starling Bank balances and transactions. Not for Xero (`xero`).
- ▌ Typecase · howells bundleDesign and enforce named typography roles across a UI codebase. Use when raw font utilities have multiplied or a type ramp needs consolidating. Not for visual direction (`chiaroscuro`), interface wording (`signage`), or one-off type polish.
- ▌ Fail Fast · howells bundleRemove hidden fallbacks, swallowed errors, legacy aliases and permissive defaults. Not for behaviour-preserving cleanup (`unslop`) or required compatibility.
- ▌ Fieldtest · howells bundleExercise a running web app in a real browser and return evidence-backed QA findings. Use for dogfooding, localhost review, responsive or mobile defects, console and accessibility checks, and persona walkthroughs; fix only when asked. Not for code-only review, or designing and building the UI (`chiaroscuro`).
- ▌ Rebalance · howells bundleReprioritize a project's backlog for a goal. Not implementation (`next`).
- ▌ Glm Review · howells bundleGet an independent read-only review from GLM 5.3 Flash on the Z.AI Coding Plan via OpenCode, then verify its claims before acting. Use when the user names GLM, or for a cheap bounded check: conformance to stated criteria, a contract, UI copy. Not for a hard judgement call (`fable-review`) or a codebase grade (`simplify`).
- ▌ Marginalia · howells bundleAdd concise, useful JSDoc where IDE hover help or a generated API reference needs a non-obvious contract. Use for exported JavaScript or TypeScript APIs, components, hooks, classes, complex types, or package publishing. Not for self-explanatory internal code, narrating comments, prose docs, or anything that changes behavior.
- ▌ Mastraudit · howells bundleAudit a Mastra codebase against execution failures first: workflow step size, fan-out keying, suspend and resume payloads, load-bearing writes, model settings, and visible tool keys. Use for pre-ship review or an existing Mastra implementation. Not for building features; use version-matched Mastra docs.
- ▌ Chiaroscuro · howells bundleDesign polished responsive web UI with Tailwind v4. Not for browser QA (`fieldtest`), reuse (`componentize`), copy (`signage`) or typography (`typecase`).
- ▌ Componentize · howells bundleConsolidate shared UI, split multi-responsibility files or extract packages. Not for redesign (`chiaroscuro`), typography (`typecase`) or behaviour-preserving cleanup (`unslop`).
- ▌ Fable Review · howells bundleGet an independent review from Claude Fable 5.1 through the Claude CLI, then verify its claims before acting. Use for a hard judgement call: a design or architecture decision, a taste question, a plan worth arguing with. Not for a cheap conformance check (`glm-review`), a codebase grade (`simplify`), or a routine diff.
- ▌ Web Research · howells bundleResearch the open web with Exa and Tavily and synthesize one cited answer. Not for browser QA (`fieldtest`) or scraping (`firecrawl-*`).
- ▌ Product Description · howells bundleDocument user-visible product behaviour from code, tests and the running product, then consolidate defects into triage. Use for complete experience documentation. Not for marketing copy, READMEs, single-feature specs, or browser QA alone (`fieldtest`).