Recently updated open-source projects

Robotics Agent Skills supplies focused guidance for ROS1, ROS2, testing, bringup, perception, design patterns, containers, and security. Loading the relevant SKILL.md files helps agents account for issues such as QoS compatibility, lifecycle nodes, bounded callbacks, health checks, and safe shutdown rather than generating syntax-only ROS code. The skills can be copied into Claude Code, Cursor, or custom agent frameworks.

Webflow Skills gives compatible agents guided workflows for managing CMS content, auditing links and accessibility, optimizing assets, working in Designer, and publishing changes through a preview-and-confirm process. The same plugin also covers Webflow CLI tasks and Code Component creation, deployment, and troubleshooting. Site operations require a Webflow account and configured Webflow MCP server, while CLI and component workflows additionally require Node.js and the Webflow CLI.

AgentCall lets a coding agent participate in Google Meet, Microsoft Teams, and beta Zoom calls as a meeting bot. The agent can hear live transcripts, speak through text-to-speech, inspect shared screens, exchange chat messages, and present content while retaining its coding session context. Setup requires a compatible agent runtime, Python or Node.js dependencies, and an AgentCall account.

Vox Director takes a one-line topic through beat mapping, a visual style choice, collage keyframes, motion, narration, music, captions, and ffmpeg assembly. It runs as an agent skill with Atlas Cloud models plus local ffmpeg, and pauses for approval of the narrative beats and visual direction before rendering the final MP4.

validate-idea gathers search, community, and competitor signals from Chinese and English markets, then has Claude score the results and produce a bilingual decision report. Seven fetchers fail independently, and data-poor dimensions are kept low rather than filled with invented evidence, so the comparison can expose when demand is already served in another market. Firecrawl, Zhihu, Brave, or proxy configuration is optional by source. The project supplies the data layer and uses the running Claude for scoring, so it needs no separate LLM API key.

okctl is the command-line client for the Ownkube developer platform. It handles browser-based authentication, AWS account connection and verification, and inspection of clusters, environments, registries, deployments, logs, revisions, and connection details. The repository also ships an agent skill that helps Claude, Cursor, Codex, Copilot, or Cline choose the appropriate command for diagnostics and resource lookups.

Ask a medical, biological, psychological, or health question in everyday language. The skill converts it into scientific search terms, asks for clarification when needed, and queries PubMed, Semantic Scholar, bioRxiv or medRxiv, ClinicalTrials.gov, and OpenAlex. Results include study types, abstracts, source links, and an evidence synthesis, using public APIs that require no keys.

Baseline is a zero-dependency readiness checker for repositories. It produces a scorecard covering build and tests, security and supply chain, reproducibility, operability, change governance, community and documentation, and claims discipline; blockers can fail CI while judgment calls go to a dated sign-off ledger. It can run as a Claude Code skill or directly through its Node CLI.

verified-experiments adds a Python harness and Claude skill with guards for data and Git provenance, real-data sanity, label leakage, impossible metrics, stored predictions, reproducibility, silent fallbacks, guard meta-tests, and code review. Each guard rejects a deliberate fake, and the diagnoser explains the failed gate and next step. An optional agent loop may edit experiment code but cannot change the guards, tests, reviewer, or diagnoser to make a run green.

awesome-claude-skills curates Claude skills, resources, and tools for document work, development, data, communication, media, project collaboration, and security, with guidance for creating custom skills. It also highlights the connect-apps plugin, which uses Composio to connect Claude to more than 500 apps for actions such as email, issues, and Slack posts. That connector needs a Composio API key, and listed projects have their own setup and licenses.

apple-hig reviews interfaces for iOS, iPadOS, macOS, watchOS, tvOS, and visionOS against Apple Human Interface Guidelines. It looks at platform-specific interaction, accessibility, design tokens, and SwiftUI code, then reports actionable issues; the project lists Claude Code and an internet connection for current guideline data as requirements.

fiz-invoicing-skill uses the FIZ API to find or create customers and products or services, apply the appropriate VAT rate, and build a draft invoice with discounts or payment details. You then issue the document—the legally binding step reported to the Portuguese tax authority—and download its PDF. Claude Code, a FIZ account, and an API key are required.

superstack guides non-coders from an idea through scope questions, a project roadmap, file generation, tests, and a local preview they can click through. After approval, it can save the files or deploy a site and can continue updating an existing project; the listed requirements include Windows, a Claude account, and a stable internet connection.

Dashmotion converts a plain-language workflow or a supported Mermaid diagram into one self-contained animated HTML/SVG file. Flow mode shows execution moving through stages and branches, while architecture mode adds component boundaries and animated request journeys. The output is small, dependency-free, editable through follow-up natural-language requests, and accessible with pause controls and reduced-motion handling. Sequence, class, ER, and gantt diagrams are outside the supported input set.

spatiotemporal-critique reviews code or writing by first preserving what already works and clarifying the intended target. Its spatial pass examines the current artifact, while its temporal pass checks edit history for regressions, drift, and changes that introduced a problem. An independent panel can add a local Ollama model, Codex CLI, or cloud model from another family. The core skill uses Markdown and standard-library Python; optional critics need their own runtime or key.

context-graph-compressor reads a standard JSON chat export, extracts topics and decisions into a weighted logical graph, and writes a smaller file that another AI tool can load to resume the work. It runs locally without extra developer tools and keeps the chat on the machine. The current release targets 64-bit Windows 10 and 11.

This Windows application helps site owners configure AI-search metadata, llms.txt, MCP, and OAuth options, then verifies that the resulting files are present and readable. A report view logs access attempts and content settings control exposed pages; it requires Windows 10/11, internet access, and administrator rights, while search visibility remains dependent on external crawling.

anti-ai-writing scans English and Spanish drafts for recurring buzzwords, empty phrases, rigid sentence structures, em dashes, and other patterns associated with machine-like prose. It returns a rewritten version plus a diagnosis explaining the changes, and supports copying the result for further editing. The published app runs locally on Windows, so pasted text stays on the machine rather than being sent to a cloud service.