Recently updated open-source projects

Use this Claude Code skill to turn a technical or regulatory topic into one self-contained HTML explainer for both executives and practitioners. It supplies content structure for audience switching, diagrams, comparison tables, glossaries, regulatory callouts, and primary-source citations, and can validate output with Playwright screenshots and a linter. The page itself is standalone, while the bundled validation tooling requires Node dependencies and Chromium.

Tribunal reviews only the current branch diff against a base branch. A hater examines each changed file, an integration reviewer looks for broken cross-module contracts, and a judge tests whether each accusation is deliberate or defensible before the verdict keeps the real weak spots. The skill writes a severity-ranked report to docs/reviews/ and works across common languages without external dependencies. It is not a general audit of the repository’s older, unchanged code.

cloud-sandbox-vm-skills teaches agents when and how to use Box’s cloud Linux VMs for persistent servers, isolated code execution, untrusted workloads, Docker, parallel agents, tunnels, public URLs, background workers, browser control, and ephemeral CI. The skills drive Box through its JSON CLI and target tasks that benefit from persistent storage, SSH, and real networking. Box is EU-only today and is not the best fit for GPUs, sub-500ms cold starts, very large VM concurrency, or process-level forks.

publishing-skills derives a public release repository from a private agent-skill development folder. It stages a copy, removes machine paths and private references, scans for leaks, adds a demo-forward README, license, manifests, changelog, and installable plugin shape, then runs verification gates. The source folder is never mutated, and the workflow stops before pushing because publication is irreversible. The result remains usable as plain SKILL.md content across providers, with Claude Code plugin support as an extra path.

Tochi-Satei-Kun reads MLIT transaction data and public-land-price GeoJSON, re-estimates municipality-specific hedonic regression coefficients, selects comparable cases, and exports broker and client worksheets in Excel. The broker sheet exposes coefficients and adjustments so the estimate can be checked, while the client sheet hides statistical detail. It supports Japanese vacant land with ownership rights for preliminary brokerage screening, not a statutory appraisal, and needs current source data supplied by the user.

swarmskill-agent-skill documents how a compatible agent can use SwarmSkill, a hosted multi-agent token-trading service registered on Ethereum. The service coordinates votes to buy a Solana pump.fun coin, sets a holding duration, sells together, and distributes payouts under its stated rules; joining uses a wallet signature and a minimum contribution, with a different share for NFT holders. This repository contains only the usage guide, while the service remains proprietary and online.

Run one command to turn a mixed track into instrumental, full-vocal, harmony, lead-vocal, extracted-reverb, dereverbed, and lightly de-essed outputs. The cross-platform Python driver uses several audio-separator model ensembles and can also be invoked as an agent skill. It requires Python, ffmpeg, and audio-separator, downloads roughly 5–6 GB of models on first use, and runs much more slowly without acceleration. The downloaded separation models have their own licenses, which should be reviewed separately for commercial use.

java-springboot-code-review examines Java and Spring Boot code beyond style and syntax. It traces transaction and AOP proxy pitfalls, JPA query and locking problems, race conditions, idempotency, IDOR, payment and inventory logic, API contract issues, and data exposure. Each review organizes critical issues, ripple effects, warnings, suggestions, and a pre-merge checklist for engineers working on high-impact services.

Use this Claude Code skill to build single-file HTML pages with a defined type scale, spacing system, one accent color, inline icons, and a light theme with dark-mode toggle. Its validators render mobile, tablet, and desktop screenshots with Playwright and run static and accessibility checks. The skill repo needs Playwright, Sharp, and Chromium installed for the full validation workflow; consumer projects receive the generated page, not copied tooling.

Shape an academic manuscript around the conventions that apply to its paper type and intended venue, rather than applying one generic writing style. The skill asks for those two inputs, then guides section structure, argument moves, voice, citation format, LaTeX choices, and reviewer-risk checks across six paper types and five venue categories. It does not invent citations, search the literature, verify facts, create figures, or run statistical analysis.

Give Find Angles an unfocused topic and it produces a structured tree of distinct research directions, concrete sub-angles, keywords, and an academic search query for every node. The workflow stays anchored to the terms you supplied and can run through Claude, Codex, or another capable model without web access. It maps where to investigate; it does not find papers or claim that proposed findings already exist.

This skill reviews one index.html for factual accuracy, clarity, professional prose, and structural quality against its referenced patterns. It returns severity-ranked findings and walks through them one at a time, applying only approved edits back to index.html; it is a post-authoring QA pass rather than a live writing linter.

codex-council is a Claude Code plugin that sends staged context to two to six parallel OpenAI Codex sub-agents, assigns task-specific roles, and aggregates their replies. Each role keeps its own thread per project and host session, and both Claude Code and the Codex CLI must be authenticated. The plugin launches Codex with approvals and sandboxing bypassed, giving each sub-agent broad filesystem access, so it should not be used with untrusted projects or content.

Timeln's skills combine six memory-recall actions with a human-gated consulting workflow. Thinking-OS retrieves decisions, failures, shipped artifacts, and call material through Timeln MCP, while Consulting-OS moves from framing and solution options through gates, acceptance, commercial terms, packaging, integrity checks, red-team review, and pursuit. The skills require a Timeln account and MCP token for hosted memory, and they state that every answer cites a source or plainly reports no record; the podcast tool can use a local TTS engine.

Add one marketplace to browse and install individual skills for Claude Code, Copilot CLI, Codex, Gemini CLI, OpenCode, Cursor, and Windsurf. External repositories are pinned to reviewed commit SHAs, and updates require another review before entering the catalog. The current marketplace contains one prompt-engineering entry, so the catalog is small even though the format supports more skills.

Manage each Slidev presentation as its own repository: scaffold it, define a palette, run review rounds, polish layout, build, and export PDF, PPTX, or PNG. Approval requires at least two review rounds, and the generated GitHub Actions workflow publishes to Pages on pushes. Setup needs GitHub CLI authentication, Node, and a configured destination directory.

nexus-mod-publisher prepares a GitHub mod release for Nexus Mods by drafting a player-facing changelog from tags, converting Markdown to BBCode, linting the archive, and mapping the actual upload or update form field by field. Its dependency-free Python scripts can run separately and remove common macOS archive junk while flagging misplaced DLLs or debug symbols. Because Nexus has no upload API, a connected browser can guide the form, but the final Publish click remains with the user.

breakdown-strat convenes five isolated advisors around a costly decision, covering leverage, cash flow, demand, minimalist entrepreneurship, and ADHD execution patterns before a chair maps disagreements into one irreversible action for the next 24 hours. A decision log blocks near-duplicate rumination, while a context file supplies the user’s real constraints and missing decisive facts stop the run. The clinician persona stays in its lane and the project is not medical advice.