{ "markdown": "
\n\n# Agent Skills\n\n**UI audits, typography, docs, PR review, and releases, loaded on demand by your coding agent**\n\n26 skills for the parts of shipping that code review never covers: the loading states, the type scale, and whether half the diff is AI slop.\n\n

\n \n \n \n \n \n \n

\n\n
\n\n## Install\n\n```bash\nnpx skills add mblode/agent-skills -g --agent claude-code codex cursor -y\n```\n\n`--agent` takes a space-separated list. A comma-separated one is rejected whole as a single invalid name.\n\nThe bundle is listed at [skills.sh](https://www.skills.sh/mblode/agent-skills).\n\nAgents: these skills. Humans: [Taste Training](https://blode.co/taste-training), a course on spotting and fixing AI slop. First unit free.\n\n### Cloud agents\n\nA cloud agent runs on a fresh VM that never sees your machine's `~/.claude/skills` or `~/.agents/skills`, and Cursor's Sync Skills toggle covers only `~/.cursor/skills`. Nothing you install locally reaches them.\n\n**Claude Code cloud, Cowork, and routines.** This repo is a plugin marketplace. On claude.ai, open **Customize > Plugins > Add > Add marketplace > Add from a repository**, enter `mblode/agent-skills`, and install the Agent skills plugin. Every cloud session gets all the skills with no per-repo config, and Sync automatically keeps them current whenever the repo changes on GitHub.\n\nTo scope it to one repository instead, declare the marketplace there and Claude installs it at session start:\n\n```jsonc\n// .claude/settings.json\n{\n \"extraKnownMarketplaces\": {\n \"mblode-agent-skills\": {\n \"source\": { \"source\": \"github\", \"repo\": \"mblode/agent-skills\" }\n }\n },\n \"enabledPlugins\": { \"agent-skills@mblode-agent-skills\": true }\n}\n```\n\n**Cursor cloud agents.** Cursor has no account-level equivalent, so install on the VM at build time:\n\n```jsonc\n// .cursor/environment.json\n{ \"install\": \"npx skills add mblode/agent-skills -g --agent cursor -y\" }\n```\n\nThat lands in the VM's `~/.agents/skills`, which Cursor loads, and leaves nothing untracked in the agent's diff.\n\n## Skills\n\n### Architecture\n\n- **[codebase-architecture](./skills/codebase-architecture/SKILL.md)**: Design a structure, deepen an existing one, or harden it with guardrails.\n- **[scaffold-nextjs](./skills/scaffold-nextjs/SKILL.md)**: Next.js turborepo with Blode UI, Ultracite, GitHub, and Vercel.\n- **[scaffold-cli](./skills/scaffold-cli/SKILL.md)**: TypeScript CLI package: ESM, tsdown, vitest, changesets, CI.\n- **[multi-tenant-architecture](./skills/multi-tenant-architecture/SKILL.md)**: Tenant isolation, routing, and custom domains on Cloudflare or Vercel.\n\n### Design\n\n\n- **[product-design](./skills/product-design/SKILL.md)**: What the interface should do, before anyone builds it.\n- **[ui-design](./skills/ui-design/SKILL.md)**: Visual direction, design-system extraction, Tailwind builds, screenshot to markup, dark mode, responsive, and a React and Next.js UX audit with a ship verdict.\n- **[ui-verification](./skills/ui-verification/SKILL.md)**: Boots the app in a headless browser and runs nine probes that turn inferred UI defects into measured ones, then proves the fix cleared.\n- **[ui-animation](./skills/ui-animation/SKILL.md)**: Springs, gestures, easing, and curves pulled from a screen recording.\n- **[presentation-creator](./skills/presentation-creator/SKILL.md)**: Decks with a story spine, speaker notes, and a contrast-checked QA pass, as Marp markdown, a web app, or a handoff to pptx.\n\n### Writing\n\n- **[eli5](./skills/eli5/SKILL.md)**: Session talk in plain language: optional analogy, exact technical terms, and house vocabulary without forced templates.\n- **[docs-writing](./skills/docs-writing/SKILL.md)**: Diataxis doc types, 51 rules, audit and writing modes, agent-readable docs.\n- **[readme-creator](./skills/readme-creator/SKILL.md)**: Type-aware README from the manifests: one install, a runnable quickstart, badges only where published.\n- **[seo-program](./skills/seo-program/SKILL.md)**: Keyword and prompt-volume research, writer briefs, AEO question maps, Search Console monitoring.\n\nPersonal voice, blog posts, and voice evaluation live in [ghostwriter](https://github.com/mblode/ghostwriter).\n\n### Quality\n\n- **[ax-audit](./skills/ax-audit/SKILL.md)**: Agentic experience audit: 27 rules for tool parity, approval gates, and escape hatches, ship verdict.\n- **[dx-audit](./skills/dx-audit/SKILL.md)**: Libraries, CLIs, SDKs, npm packages: 38 rules, agent-friendly checks, root-cause findings.\n- **[typography-audit](./skills/typography-audit/SKILL.md)**: 78 rules: punctuation, fonts, sizing, spacing, hierarchy, pairing.\n- **[optimise-seo](./skills/optimise-seo/SKILL.md)**: Next.js App Router SEO: sitemaps, metadata, structured data, redirects, hreflang, AI-crawler policy, Core Web Vitals.\n\n### Shipping\n\n- **[planning](./skills/planning/SKILL.md)**: Writes and reviews executable plans with repository evidence, vertical slices, and explicit acceptance criteria.\n- **[pr-reviewer](./skills/pr-reviewer/SKILL.md)**: Read-only diff review: bugs, structure, AI slop, security.\n- **[tidy](./skills/tidy/SKILL.md)**: Applies diff-scoped simplifications, preserving necessary guards and existing edits.\n- **[pr-creator](./skills/pr-creator/SKILL.md)**: PRs with short human descriptions, Linear IDs, templates, drafts, tidied commits.\n- **[pr-babysitter](./skills/pr-babysitter/SKILL.md)**: Watches an open PR: conflicts, CI, comments. Fixes what it can.\n- **[autoship](./skills/autoship/SKILL.md)**: npm releases with changesets: fix loop, CI watch, Version Packages merge, OIDC publish verify.\n\n### Authoring\n\n- **[agents-md](./skills/agents-md/SKILL.md)**: Wires a repo so Claude Code, Codex and Cursor read the same instructions, then audits and refactors AGENTS.md and CLAUDE.md, with grades.\n- **[agent-skills-creator](./skills/agent-skills-creator/SKILL.md)**: Creates and audits skills by their added value, with portable workflows, a validator, and regression scenarios.\n- **[save-md](./skills/save-md/SKILL.md)**: Writes a named source as a markdown file the next turn can reread.\n\n## Maintaining the collection\n\nSkills retain house preferences, operational contracts, and repeatable audit rubrics. Generic coaching belongs to the model and host. Every skill includes authored regression scenarios; these are not claims of measured improvement across models.\n\nRun `skills/agent-skills-creator/scripts/validate.sh --all` for portable format requirements and separate house conventions covering references, counts, and scenario structure. Authoring follows the [Agent Skills specification](https://agentskills.io/specification) and [best practices](https://agentskills.io/skill-creation/best-practices). The [collection audit](./maintenance/2026-09-05-skill-audit.md) records changes, retained value, and verification limits.\n\nThe canonical validator also serves `agent-evals` and the private collection. Machine output uses\n`--format tsv` (skill, status, section, check, detail); `--policy private --repo PATH` checks\nportable metadata, catalogue membership, and authored cases without imposing public path rules.\nRun `python3 -m unittest discover -s maintenance/tests` after changing this interface.\nCI validates all public skills and protocol fixtures on each pull request.\n\nUse the sibling eval CLI to check routing coverage and compare full installed folders:\n\n```bash\nnode ../agent-evals/dist/cli.js validate-cases --skills-dir \"$PWD\"\nnode ../agent-evals/dist/cli.js doctor --source \"$PWD\" --source ../agent-skills-private\n```\n\nThe doctor reports drift and retired ownership without overwriting installed edits. Reconcile\nthose edits into source before reinstalling. Behavioral comparisons must use fresh equivalent\nfixtures and report baseline results, host/model, loaded files, and unmeasured assertions.\n\n## License\n\nMIT\n\n---\n\nCrafted by [](https://blode.co) [Matthew Blode](https://blode.co)\n\n## Company identity and copy\n\nBranding, copywriting, and product judgment now live in the separate [Brandwriter repository](https://github.com/mblode/brandwriter). It maintains company identity kits and reads company profiles independently of personal Ghostwriter profiles. Install with `npx skills add mblode/brandwriter --agent codex claude-code -y` (repository access required).\n", "bytes": 8740, "sha": "07be381676dc281eb46636f6087ad28890d00e34277db63ed5100ceee27c4ce6", "repo_slug": "mblode/agent-skills", "fonte": "repo", "truncated": false, "api": "https://agentalog.com/api/listings/skl_mblode_agent_skills_ui_animation_0b83ccb4/readme" }