Shape agent capabilities into your workflow.

report-kit helps an author produce a single HTML report that recipients can open in Edge or Chrome without installing anything. The page supports in-place editing, saving changes through the browser File System Access API, and PNG export through an embedded html2canvas copy, so it works without a CDN. The author needs Claude Code and Python for bundling; Firefox can open the file but falls back to downloads for save and image actions.

voz treats writing as craft rather than as a detector or score. It removes clusters of machine-like vocabulary, symmetry, inflated phrasing, and formatting habits, then applies one of nine audience registers such as technical, journalist, legal, teacher, or marketing. You can invoke it per task or wire the default voice into Claude Code through a project instruction and SessionStart hook. Its word list is dated and a single tell is never treated as proof.

professional-repo-docs bootstraps a GitHub repository’s documentation from a natural-language request, including README variants, a suitable license, community-health files, issue and pull-request templates, AGENTS.md, and llms.txt. It detects the project stack to fill real build and test commands, copies canonical license and conduct text, and creates only missing files. The Node-based engine never auto-commits; the optional GitHub CLI is only needed to enable Discussions.

blockpad-documents helps Claude Code create or edit `.bpad` files containing units-aware equations, tables, two-dimensional drawings, and solver sections that recompute in Blockpad. It can also convert a Word file, PDF, Markdown document, or photographed hand calculation into a live calculation sheet. A Python validator checks XML structure, schema details, and known formatting or formula traps; the integration targets Blockpad 1.8.x and is unofficial, so the desktop app remains the final visual check.

Google Ads Custom Affinity Builder turns a product, market, persona, and user-supplied sub-themes into a multi-tab XLSX with keyword, website, video, channel, and app signals. It uses yt-dlp for view-aware video search, enforces a 10K-view floor, checks a channel’s recent uploads rather than trusting one viral video, and sends candidates to an independent checker. The workbook is designed for Google Ads entry, with tab names constrained by Excel’s 31-character limit.

ks-deep-claim-audit turns a post, article, screenshot, or repository link into a claim-by-claim fact report. It maps each statement to source code, command output, dated URLs, or other primary material, classifies it as true, partly true, exaggerated, false, or unverified, and cross-checks the highest-risk conclusions. The workflow is read-only and refuses to execute untrusted repository code; when the underlying value cannot be obtained, it keeps the unresolved verdict instead of softening it.

Amino Deliverability Audit inspects public DNS and related trust signals for an email-sending domain, looking beyond the existence of SPF, DKIM, and DMARC records to alignment gaps, report destinations, MTA-STS coverage, and other silent failures. It combines the signals into an effort-versus-value repair order and drafts the exact records to change. The tool is read-only: it never sends mail, asks for credentials, or edits the domain.

design-to-frontend takes an existing no-build React prototype from Claude Design and moves it toward a dev-ready frontend. It maps pages, components, tokens, and missing assets, can generate transparent images with OpenAI, then adds features and polishes hierarchy, interaction states, mobile behavior, localization, and accessibility. The workflow verifies edits in a real browser with fresh cache-busting ports. It is not a starter for a new Next.js, Vite, Vue, backend, or SEO project.

napkin reduces a sprawling strategy to three to six ordered phases, draws the condition that earns each transition, and marks a flywheel only when an output genuinely feeds back into the plan. It defaults to a compact ASCII sketch and can emit Mermaid or SVG through a small Node renderer, keeping the reasoning cheaper than repeated image renders. The skill works across many coding agents and refuses to invent a missing gate or pretend that a linear plan has a flywheel.

resend-skill helps an agent send transactional email, manage batches, verify sending domains, and check delivery status through Resend. It sets up DKIM, SPF, and a separate return-path, then can pair Cloudflare Email Routing with Gmail send-as so an address can receive replies and send as that identity. A Resend API key is required, and inbound mail needs a Cloudflare-managed domain plus a human verification step.

connect-codex-skill wires the OpenAI Codex CLI into Claude Code as a second opinion for diff reviews, adversarial plan critiques, and delegated rescue work. Its setup installs the CLI, completes OAuth, adds the OpenAI plugin marketplace, installs the Codex plugin, and verifies the connection after a restart. The workflow exposes slash commands for review, rescue, status, results, and cancellation. It requires macOS or Linux, a recent Codex CLI, and an eligible ChatGPT subscription.

open-source-octopus-investigation starts from a person or topic and follows GitHub stars, follows, followers, and contributions to surface projects that keyword searches may miss. The skill scopes an origin, depth, and call budget, then groups findings into themed tentacles, records visited networks, and keeps unexplored leads for later runs. It can also prepare issue reports. Crawling uses the authenticated gh CLI and GitHub APIs rather than scraping pages.

business-analysis-generator researches a company and its market, uses the findings to shape an adaptive interview, and turns the result into four to ten reusable analysis prompts. Each prompt specifies the business connectors, grounding choices, useful response, failure modes, and industry adaptations, making the library suitable for consulting audits and implementation planning. The company analysis can be refreshed occasionally, while the prompts can run on a recurring cadence against live business data.

ecom-static-ads reads one product or collection page, extracts the store’s product imagery and customer reviews, researches customer language and competitor ads, and develops four distinct static DTC concepts. It then generates finished images with the real product photo as a reference and returns a campaign brief, funnel recommendations, an A/B plan, and platform-specific copy. Gemini is required for image generation; research services are optional and fall back to web search.

connect-gemini-skill makes a locally authenticated Gemini CLI available inside Claude Code through /gemini, /gemini-review, and a dispatchable Gemini subagent. It supports independent opinions, Google-grounded research, very large context analysis, and diff review without switching terminals. The installer checks or installs the CLI, verifies authentication, and places the connector files. The coding-focused CLI can analyze images but does not generate Nano Banana images, and it targets macOS or Linux.

respec-editor helps update W3C ReSpec HTML specifications with correctly nested sections, definitions, normative statements, cross-references, and stable identifiers. It leaves generated tables of contents, bibliographies, indexes, and section numbers alone, and avoids changing metadata or normative status unless explicitly permitted. The skill prepares a branch, diff, and pull-request draft, then waits for confirmation before remote Git actions. It is scoped to ReSpec documents, not generic HTML, Bikeshed, or Markdown.

things-app connects Claude to Things 3 with AppleScript so it can read lists and create, update, move, reschedule, complete, or delete tasks and projects. It also imports a Reminders list as a flat Things project and turns YAML templates into reusable projects with sections. The skill requires Claude Code and Things 3 on the same macOS machine, so it is not suitable for Cowork’s remote Linux shell. Reminders imports intentionally lose section structure.

polymarket-data turns Polymarket’s public endpoints into research-ready JSONL for market metadata, outcome prices, and deduplicated trade history, with a calibration command for comparing price buckets to realized results. It handles the browser User-Agent requirement, the two-sided rows emitted for one fill, JSON-encoded market fields, and last-trade-per-minute history. The implementation uses only Python’s standard library and needs no API key or login. It is intended for public, unauthenticated data access.