Recently updated open-source projects

AEON turns repository-based skill prompts into autonomous runs that can be scheduled, chained, and monitored from a dashboard. Results can be delivered through Telegram, Discord, Slack, or email, while each skill declares the API keys or MCP servers it needs. Setup requires Node.js 20+, an authenticated GitHub CLI, and your own repository copy; some skills depend on additional credentials.

Claude Session Miner reads local Claude Code JSONL transcripts, strips tool-call noise, and extracts durable ideas, decisions, pending items, and system facts into one JSON object per line. A per-host registry prevents duplicate processing, so it can run through launchd, cron, or systemd. Apart from its inbox and registry it only reads transcripts, and its extraction prompt explicitly excludes secrets, tokens, and personal data.

Particle Device OS contains the firmware libraries and system components used to build software for Particle hardware. Developers can use its build guides, application firmware API, debugging support, bootloader, cloud communication modules, threading code, and hardware abstraction layers. It is primarily a firmware development source tree, and changes to LGPLv3-licensed system components carry the sharing obligations described by the project.

pm-skill has Claude Code discover the problem, write a specification and plan, and wait for explicit approval before implementation. It then breaks the work into stories and coordinates specialist agents for building, quality gates, review, testing, verification, shipping, and logging. The PM skill itself does not write code, and its tracked state supports resuming interrupted work.

agentlog stores project decisions as tagged JSONL entries that remain easy to grep and version-control. Use three commands to add a decision, retrieve relevant context, or ask Claude or Gemini to surface patterns, contradictions, and missing reasoning across the history. It avoids embeddings and databases, making it a small, transparent option for decision logs of roughly hundreds rather than very large knowledge bases.

Describe an architecture task in Claude Code and the skill selects an appropriate combination of C4, ArchiMate, TOGAF, or arc42 with ADRs. It can produce diagrams, documentation, decision records, graded reviews, and enterprise models while reusing stable element IDs across artifacts. A companion linter checks cross-document consistency, including undefined IDs, broken ADR links, and orphaned diagram sources.

Moltis is a self-hosted personal-agent server written in Rust that connects to multiple model providers and stays available through the web, Telegram, Signal, Discord, Teams, and voice. Tool execution can be isolated with Docker, Apple Container, or WASM, while memory, MCP, scheduling, authentication, and an encrypted vault are built in. Release installs run as a single Rust binary on hardware you control.

This repository states that Crystal was deprecated in February 2026 and directs users to Nimbalyst for active downloads, documentation, and updates. Nimbalyst is a local-first workspace for human and agent collaboration, with editors for code, Markdown, spreadsheets, diagrams, and visual documents. It uses Git worktrees to isolate parallel coding sessions and is built on Electron.

Use this Claude Code skill to capture baseline screenshots, rerun pages after code changes, and inspect an HTML diff report when the UI shifts unexpectedly. It supports multiple viewports, masking for dynamic content, and configurable tolerances. For consistent results between macOS and Linux CI, the project recommends generating baselines in Docker.

tyler_med converts a folder of clinical journal PDFs into a two-tier Markdown wiki with study-design grouping, DOI, sample size, and data-source fields. It also exports CSV and JSON evidence tables, cleaned paper notes, references, and an incremental cache, with Obsidian-compatible links. Study design, sample size, and data source are best-effort extractions, so verify load-bearing details in the full text.

Theorist keeps a THEORY.MD at the repository root that explains the problem thesis, current system model, strategy, important discoveries, and open uncertainties. The Codex- and Claude-compatible skill rewrites the document holistically when understanding changes instead of appending a task plan, changelog, or status log. It provides a lightweight strategic memory for long-running engineering work.

journal-fit-finder extracts a manuscript’s field, methods, article type, readership, language, and constraints, then ranks journal candidates with a transparent fit rubric. It checks scope, recent precedent, indexing, APC, publisher diversity, and rejection risks, while refusing to invent acceptance rates or quartiles. Journal policies, fees, rankings, and turnaround times change, so final targets still need current primary-source verification.

Enter an address, listing URL, or coordinates, and the skill detects the country and applies the matching due-diligence playbook. You can request sections on price, tax, financing, visas, schools, scams, title monitoring, relocation, and other purchase concerns, then view the result in the terminal or save a Markdown report. Claims are dated and confidence-labeled from primarily government sources. It is decision support rather than legal, tax, or financial advice, and full audits can use substantial Claude context.

This Claude skill turns a French description of dates, times, and locations into a Framadate poll. It calculates the options, fills and publishes the form in a connected Chrome browser, then returns both the share link and the private administration link. It is useful for recurring meetings or volunteer schedules, but requires Claude desktop Cowork and the Claude in Chrome extension.

This Claude Code skill measures typography, colors, spacing, borders, shadows, and interactive states in a live or staging site through Chrome DevTools, then compares them with Figma, a PDF brandbook, a reference site, screenshots, or manual specifications. It cross-checks suspected deviations, filters false positives, and produces a structured Excel or Markdown report with severity and current versus expected values. A target URL, at least one design reference, and the Chrome MCP extension are required.

BeeAI Framework supplies Python and TypeScript components for agents, multi-agent handoffs, model backends, tools, and RAG, helping developers assemble applications that reason, act, and collaborate. Requirements can constrain agent behavior consistently across different models, while MCP, ACP, and custom tools extend the system. Examples work with hosted providers and can also run locally through Ollama.

gwdg-gipplab-skill is a Claude Code plugin and skill guide for the GWDG/NHR-Nord Göttingen clusters Emmy and Grete. It organizes account types, SSH and transfer paths, Slurm jobs and partitions, storage quotas, Lmod/CUDA/Apptainer environments, and Hugging Face caching into focused references. Install it through the plugin marketplace or use the bundled skill when working on those systems.

AuditKit fetches an authorized site’s robots.txt, sitemap, HTML, headers, and JSON-LD, then uses deterministic checks for SEO, performance signals, accessibility, conversion, security headers, structured data, and AI-search readiness. It applies a published weighted rubric and returns prioritized, copy-ready fixes, while clearly flagging limits such as backlinks, rankings, field performance data, authenticated pages, and content the environment cannot render.