Recently updated open-source projects

Lumate takes a Luma event link, reads the form, fills profile fields, and drafts answers from the user’s real GitHub projects before submitting the registration. The first event helps establish reusable answers, so later registrations can run with fewer questions. It keeps the data local, but users should still verify the drafted responses and final submission because the skill acts on an external event form.

VAP Media Skill lets compatible AI assistants start image, video, and music generation through one API, poll jobs for results, and run media editing or enhancement operations. It can be installed for Claude Code, Codex CLI, or Gemini CLI, but service calls require a VAP Media API product key.

supertutor teaches through plain definitions, concrete worked examples, and explain-back rather than accepting a learner’s nod. After each restatement it identifies the first gap using a fixed rubric and ladders hints until the learner can rebuild the idea. The skill is aimed at genuine understanding, with first-principles engineering as its flagship example; simple one-line facts are better answered directly.

delegate-local routes self-contained text jobs such as log summaries, changelog drafting, classification, extraction, and prose rewriting to an installed Ollama or MLX model. This keeps source material on the device and reduces the amount of routine output occupying the cloud agent's context, while reasoning-heavy, tool-using, and repository-wide work stays with the main model. It is most useful for repeated or high-volume tasks and requires a working local model backend.

Google Ads Keyword Classifier organizes Keyword Planner exports by search intent, including negative, competitor, branded, category, feature or problem-aware, comparison, and informational terms. It filters low-volume or irrelevant entries, assigns primary, backup, or excluded status, and samples regional SERPs to check the classification. The result is a four-sheet Excel workbook with the keyword list, counts, competitor terms, and negative keywords for campaign planning.

Personal Trainer skill combines evidence-based strength programming and nutrition guidance for one athlete. It can create RPE-based progressive plans, log sessions, track progress and plateaus, calculate calorie and macro targets, assemble practical meals, and retain the athlete profile in local files between conversations. The same knowledge base is exposed through Claude, Codex, Cursor, Gemini CLI, and other agents. It supports coaching workflows, while the athlete remains responsible for applying advice to their own circumstances.

auto-skill-manager scans the available skills, installs missing ones, and activates the right tools for a task across Claude Code and Cowork. It supports Windows, macOS, and Linux, uses Python 3.8 or newer for scanning, and can bring in optional Python, Node, or command-line dependencies when needed. The convenience comes from handling local setup instead of requiring manual installation steps.

pull-context helps a Claude Code session import context from another session on demand. It searches candidates by topic, age, and message count, lets you choose a source other than the current session, and distills the long transcript into goals, decisions, pointers, and open questions. A verbatim tail can be pulled when exact wording matters. It relies on Claude’s local transcript layout, which is undocumented and may change.

This Claude Code plugin helps developers set up Lore, run everyday version-control commands, understand its content-addressed model, and work with the Python, JavaScript, C#, and Go SDKs. It also maps common failures to practical fixes and advises checking the installed CLI help because Lore is still pre-1.0. The project is community maintained and is not endorsed by Epic Games.

This kit converts text-layer PDFs from Traditional to Simplified Chinese while trying to retain page size, images, selectable text, and overall layout. An image-only cover can be replaced in a separate step, and the workflow is also packaged as a Codex skill. The included example PDFs and cover are not automatically relicensed under the code’s MIT license.

ai-equity-analyst reads only the investor presentations and earnings-call transcripts you provide for an Indian listed company. It applies a 33-question checklist, distinguishes hard signals from soft guidance, and produces bull and bear cases with passage-level citations plus a promise-versus-reality ledger. It is for research and education, not SEBI-registered investment advice, and it does not fill gaps from memory.

home-auto-insurance-quotes connects an agent to a network of licensed independent agencies for live home, auto, and bundled insurance quotes. It supports conversational intake or an agent-driven quote-to-bind handoff, while the agency’s licensed human finalizes the policy. Coverage is focused on Texas today and eligibility is checked by state and product; paid rating and personal-data tools require a partner API key.

Andie organizes an AI advisor room around four modes: Deep explains a subject, Drama debates a contested choice, War focuses on immediate production stabilization, and Kaizen investigates recurring failures. Named experts and a Devil’s Advocate challenge the framing, while the user keeps the casting vote in decision work. It runs across several major AI platforms, but its value is the structured conversation rather than an automatic decision.

MDA is a Markdown-based specification and CLI for maintaining agent-facing documents from one source. Authors write a single .mda file and compile it into SKILL.md, AGENTS.md, MCP-SERVER.md, and CLAUDE.md outputs, with JSON Schema validation, typed relationships, dependency metadata, content digests, and signature fields. It is intended for authoring and CI checks rather than as an application runtime dependency. A bundled signature verifier, dependency resolver, and central registry are not yet included.

SnapDeck drives a running web app with headless Chrome, captures real UI states, reads element coordinates, and renders HTML/CSS slide designs to PNG before assembling a full-bleed .pptx. It can reuse a logged-in session, open tabs or modals, force light mode, and place numbered callouts beside the right controls. The resulting slides are crisp picture slides for demos or training, not editable PowerPoint text and shapes.

font-pairing-skill helps an agent choose display and body faces, match typography to a project palette and icon weight, and build a coherent type scale. It includes eight archetype recipes, variable-font guidance, Google Fonts import references, and loading patterns such as preconnect, preload, font-display, and subsetting. The goal is a considered type system that avoids layout shift rather than a random font list.

icon-system-skill guides library selection across Lucide, Phosphor, Heroicons, Radix Icons, and React Icons, then defines optical sizing, stroke-versus-fill choices, weight matching, and color inheritance through currentColor. It also covers animated SVG patterns, aria labels, and 44×44px touch targets, with eight project-archetype recipes. Use it when an interface needs a consistent icon language rather than one-off symbols.

psycho-cybernetics gives Claude a mindset lens based on self-image, clear targets, and feedback. It encourages the user to aim, act, read failure as steering data, and adjust instead of treating setbacks as a verdict about identity. Invoke it with /psycho-cybernetics or let it trigger when the obstacle is mental rather than informational. It is a coaching frame, not a diagnostic tool.