Recently updated open-source projects

html-to-pdf-agent-skill uses Playwright to render remote HTML pages as local PDFs. It can handle one page or wait for a single-page application to load, extract real documentation links from the rendered sidebar, convert each page, and merge the results. Page size, margins, styles, and authentication can be customized; Playwright with Chromium is required, and full-site merging additionally depends on PyPDF2.

claude-skill-mercadopago-api gives Go developers an end-to-end MercadoPago Checkout Pro workflow for creating preferences, redirecting buyers, receiving webhooks, validating HMAC signatures, and confirming payment status. It includes ready-to-adapt Go code, status-detail mappings, Argentine test cards, and troubleshooting references. The documentation is primarily Spanish and focuses on Argentina, with separate credentials required for sandbox and production.

mac-proxy-sync reads the active macOS proxy from scutil and exports common HTTP, HTTPS, SOCKS, and no-proxy variables whenever zsh starts. A prompt hook refreshes the values at most once per minute so long-running sessions follow rotating Clash, V2Ray, or Mihomo ports, while syncproxy mirrors them into launchctl for GUI subprocesses. It also unsets stale values when the system proxy is disabled and is specific to macOS and zsh.

Nano Banana 2 Skill connects Claude Code to Google's Gemini 3.1 Flash Image model for text-to-image generation, natural-language editing, and multi-image composition. It supports resolution and aspect-ratio controls, iterative refinement, search-grounded prompts, and up to 14 reference images. Users must provide a Gemini API key and install the included Python dependencies.

Vingo embeds technical-English practice into conversations with Claude for Vietnamese speakers. It explains unfamiliar terms in everyday Vietnamese, corrects pronunciation and grammar, and uses escalating comedic roasts to make mistakes memorable, targeting beginners, founders, product or design roles, and developers who want clearer spoken tech vocabulary.

Floor Report Skills serves industrial maintenance and reliability leaders. Its current plant-pulse-report skill converts maintenance KPIs into an executive summary, trend-backed KPI cards, three priority actions, and a PDF-ready report. The repository also includes a 90-day maintenance improvement field guide, with its benchmarks and recommendations tied to named maintenance standards and reference works.

Provide a topic, an optional research file, and an iteration count. Each fresh subagent reads the document's findings, investigated paths, decision log, and open questions, then spends its rounds only on unresolved gaps before updating the same file. Dead ends are marked `UNFILLABLE`, and the document is the sole handoff between iterations, preventing later passes from rediscovering already covered facts.

Start from a brand configuration to build animated HTML5 banners across more than 15 standard IAB sizes, render backup images and previews, package network-specific ZIP files, and validate common ad-operations requirements. Current outputs target Google Ads, DV360, The Trade Desk, and Adform. The project is still alpha; its config-driven pipeline is available, while Figma MCP ingestion and additional social formats remain planned.

nadicode-ai-designer teaches coding agents to route icons, illustrations, logos, hero images, Lottie animations, and short clips to an appropriate generation path. Agents can author simple SVG assets directly or run bundled scripts for several image and video providers, all returning a consistent JSON result. Most hosted providers require their own API credentials, while the agent-native SVG path can run without an external call.

routine-writer teaches agents to express inbox triage, pull-request reviews, alert checks, briefings, trackers, and fixers as terse routines rather than long operational playbooks. It defines a title, tagline, and body pattern with verb-first instructions, explicit iteration, optional actions, empty-state handling, and a short report. The skill follows the Agent Skills specification and can be installed in compatible harnesses.

Skill Security Auditor checks skill files and MCP servers for dangerous commands, prompt injection, credential access, risky tool combinations, supply-chain issues, and MCP-specific flaws such as SSRF or path traversal. It produces a heuristic risk score and report, but its static analysis can miss sophisticated behavior and should be paired with expert inspection.

Provide a site URL, business summary, audience, and growth goal to receive a topic-cluster plan tailored to the site's type and maturity. Topic Cluster Architect maps pillar pages, supporting content, internal links, and priorities, and can use competitor URLs or an existing content inventory to identify gaps. The result is a phased execution roadmap rather than a disconnected keyword list.

Ritual scans shell history, Git repositories, Claude Code memory, and existing automations on a Mac to rank the five repeated workflows most worth automating. It then drafts a paste-ready routine prompt for the top pattern using your actual repository names and working context. The result is intended for Claude Code scheduled triggers, giving users a concrete first automation instead of a blank scheduler.

Hotspot Radar helps Chinese content creators choose timely topics by fetching top items from Weibo and Baidu and estimating keyword interest from Baidu suggestions. It runs as a Python 3.8+ script with custom keywords, result limits, and JSON output, without requiring an API key or account.

ChronoPlanner turns wake time, chronotype, sleep quality, task priorities, and fixed commitments into an annotated day or week plan. It can pre-scan connected calendars, task lists, email, and other tools, then adjust only affected blocks when plans change and learn from end-of-day reviews. It is designed for flexible schedules rather than fixed 9-to-5 routines or long-term project management.

Give the skill a market topic and optional business context, and it gathers material from nine source types through Claude Code’s web tools and the GitHub CLI. It extracts signals, clusters themes, builds a Playing to Win strategy with inline citations and falsifiable assumptions, then adds objective and expert-style scores. The result is a self-contained HTML report with linked evidence, while saved stage files make an interrupted run resumable.

Find the page you were viewing around a meeting timestamp, search recent titles and URLs, list visits to a domain, or inspect activity across browser profiles. The single-file Python tool copies each local SQLite history database to a temporary location and queries the copy read-only, with no network calls or access to passwords, cookies, or form data. It supports Chrome, Edge, and Firefox, but not Safari.

Use two Claude Code skills to research, build, update, verify, audit, scan, and maintain a body of content through staged multi-agent workflows with human checkpoints and resume support. Domain-specific entities, claim types, source rankings, paths, and integrity rules live in replaceable JSON configurations, leaving the orchestration reusable across subjects. Markdown, Laravel, and Next.js outputs are supported; generic headless CMS delivery is still a placeholder.