Recently updated open-source projects

XDeck is a macOS client for X that recreates a TweetDeck-style multi-column workspace. Users can place following, For You, notifications, profiles, or custom list URLs side by side and adjust the layout through a settings file. It includes ad blocking and can be installed from GitHub Releases or Homebrew, making it most useful to people who prefer monitoring several timelines at once.

ClaudePrism is a local-first desktop workspace for scientific writing. Users can start papers, theses, posters, presentations, and other LaTeX projects from templates, edit them locally, and compile with the embedded Tectonic engine. It also integrates uv for project-level Python environments, making plots and data analysis available inside the same workflow, and can install domain-specific scientific skills. Files are stored and compiled on the machine, but prompts and any content read by Claude are sent to Anthropic's API when AI features are used.

Kaya gives Go players a browser and desktop workspace for playing games, editing variations, organizing SGF records, and reviewing positions. KataGo analysis adds win rates, move suggestions, quality graphs, and ownership heatmaps, while photo recognition can turn a physical board into an editable SGF. It runs on Windows, macOS, Linux, the web, and as an installable offline-capable PWA, making it suitable for both game review and regular study.

Swift Concurrency Pro is an agent skill for Claude Code, Codex, Cursor, Gemini, and other coding assistants. It guides writing and review across async/await, actors, Sendable, task groups, structured concurrency, newer language features, and migration choices, with rules aimed at mistakes language models commonly make. After installation, users can invoke the skill with a dedicated command or natural-language request and narrow the review to a specific concern. It supplements an existing coding agent rather than running as a separate Swift development tool.

Groove Basin runs a music server connected to a personal library and lets laptops, tablets, and phones control playback through a web interface. It includes automatic queueing, drag-and-drop uploads and playlists, loudness scanning, filesystem monitoring, remote streaming, and MPD compatibility. The project is for music you own or manage locally and does not support DRM content.

Imagine is a desktop image optimizer for preparing graphics for the web or reducing everyday asset sizes. Its graphical interface can batch-compress PNG, JPEG, and WebP files and convert between supported formats. Prebuilt releases are available for Windows, macOS, and Linux, making it suitable for users who prefer not to assemble a command-line compression workflow.

This reference architecture shows how to build a Strands Agents chatbot on Amazon Bedrock AgentCore with persistent memory, MCP-connected tools, code execution, browser automation, voice interaction, and A2A delegation to specialist agents. Terraform modules and a working frontend/backend structure make it useful for teams evaluating or extending advanced agent workflows on AWS.

Story Skills gives writing agents a shared Markdown project format for planning and drafting fiction. Story bibles, characters, worldbuilding, factions, artifacts, plot arcs, timelines, scene state, promises, payoffs, and chapter drafts are kept as plain files with YAML frontmatter. Compatible agents handle the creative workflow, while an optional companion CLI initializes projects and performs deterministic continuity checks, such as detecting dead characters returning or payoffs appearing before setup. The bundle supports Codex, Claude Code, and other Agent Skills environments.

Bulletproof is a development methodology packaged as an agent skill for Claude Code, Codex, Cursor, Gemini CLI, Windsurf, and OpenCode. It makes the agent research the system, define a specification and acceptance criteria, plan, build, audit its work, verify behavior and impact, and prepare a handoff instead of jumping straight into code. The workflow adapts to task size: small fixes use a lightweight subset, normal features use most stages, and architecture changes use all twelve. Templates and a review sub-agent support consistent use.

This repository packages Agnes free-model access as three independently installable skills for text, images, and video. Local Python helpers cover chat and streaming responses, image generation and transformation, and asynchronous video creation with status polling and downloads. An Agnes API key must be supplied through environment variables, while the repository deliberately excludes real credentials, generated media, and local outputs. Users can install only the capability they need.

SkVM profiles a model-and-agent-runtime combination, compiles a skill against those capabilities, and then evaluates and improves it through precompilation and runtime-optimization workflows. It is aimed at teams that need the same skill to behave across heterogeneous agents rather than tuning prompts by hand for each runtime. Provider configuration is required, and the bundled benchmark suite lives in a separate data submodule.

Heptabase CLI Skills gives skills-compatible agents a structured way to manage notes, journals, tags, cards, files, whiteboards, goals, courses, and lessons through the Heptabase command line. The package can be installed for Claude Code, Codex CLI, OpenCode, or other implementations of the Agent Skills specification. It is a skill layer for an existing CLI rather than a standalone Heptabase client, and should be refreshed when the CLI interface or declared compatibility range changes.

SSH Keys Manager replaces routine terminal edits with a native Mac interface for browsing, generating, renaming, duplicating, and removing key files. It also provides a form-based SSH config editor for Host blocks and IdentityFile mapping, plus menu-bar copying and reachability checks. A read-only mode blocks edits, overwrites, renames, and deletions when you only want to inspect a workspace. Current release builds are unsigned and not notarized, so macOS may require a manual quarantine workaround.

UPskill is a command-line toolkit for creating standard Agent Skills from task descriptions, examples, existing skills, or recorded agent traces. It can generate tests, compare baseline and skill-assisted performance, refine weak skills from failures, and benchmark multiple models across repeated runs. Evaluations can run locally or through Hugging Face Jobs, making it useful for teams that want measurable evidence before distributing a skill.

xint is a command-line tool for researching X/Twitter without leaving the terminal. It can search posts, watch topics over time, inspect profiles and threads, compare follower changes, and export results as JSONL or CSV for further analysis. The same capabilities can be exposed to agents through its skill and MCP interfaces. Core access requires an X API bearer token, while xAI credentials are optional for AI-assisted analysis and reports.

How to Contribute to Open Source is a multilingual resource list for people making their first contributions. It gathers introductory articles, GitHub search approaches, version-control material, books, community initiatives, and programs that welcome contributors. The repository is an educational index rather than an automation tool; readers can suggest additional resources through pull requests when they find useful omissions.

AI Runner combines an offline conversational companion with configurable personality, voice, and memory, plus a layered canvas for generating and editing images. It can also run headlessly and expose local LLM, image, speech-to-text, and text-to-speech functions over HTTP. Docker is the recommended setup, while manual installation requires Python 3.13 and a substantial set of system dependencies.

Claude Agent Team Manager turns existing Claude Code agents and skills into a visual, drag-and-drop organization chart. Users can save team configurations, share variables, generate a deployment primer from a short objective, and launch the prepared run in a terminal. Teams and chained pipelines can also be scheduled through the operating system; the macOS build is unsigned and needs manual approval on first launch.