Git archive

Repositories 962Checked 2026-11-13

8652026-10-30
SimoneAvogadro/android-reverse-engineering-skill
What it does
This repository packages Android application analysis as a Claude Code skill with local helper scripts. It is intended for analysts who have an Android binary but need to understand its structure and HTTP interfaces without the original source. Inputs are APK, XAPK, JAR or AAR files, and outputs include decompiled sources, an architecture summary, endpoint documentation and call-flow descriptions. The workflow first identifies the application framework and then delegates native Java or Kotlin decompilation to jadx or optional alternative tools. Its API search script matches source patterns for Retrofit, OkHttp, Ktor and other stacks and returns candidate locations for review. Additional helpers use surviving Kotlin metadata to map some obfuscated names back to readable names, but search matches and recovery claims do not establish complete API coverage. Java JDK 17 or later and jadx are required, and Windows PowerShell support is described as experimental. Local scripts do not make the agent's analysis private by themselves, since source read by Claude is processed under the model provider's terms, and use must remain within an authorized scope.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository
8642026-10-30
skyzh/tiny-llm
What it does
tiny-llm is a hands-on course for systems engineers who want to understand local LLM inference rather than treat a serving engine as a black box. It takes Qwen3 model weights and tokenized conversations through attention, sampling and repeated decoding to produce text. Learners implement operators with MLX arrays, Python, C++ and Metal, while MLX provides a correctness reference and performance baseline. Later chapters connect KV caches, which retain past attention keys and values, to continuous batching and paged memory management. Week 4 adds a teaching coding agent that validates JSON actions, obtains approval for effects and records observable results. The learner package contains unfinished exercises and is separate from the reference implementation. The supported native environment is an Apple Silicon Mac with the specified Python dependencies, Xcode and native extension toolchain. This is teaching material rather than a production sandbox or durable transaction system, and the real-model agent CLI does not compose all the later checkpoint and branching mechanisms.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository Website
8632026-10-30
LOWERTOP/Shadowrocket
What it does
This repository is a community manual and annotated configuration collection for Shadowrocket, a network application for Apple devices, rather than the application's source code. It helps users understand server entries, routing rules, DNS settings and connection failures. Users read the manual or search its published website, then enter their own server details and configuration into the separately obtained application. The examples describe how rules select direct connections or proxies and how proxy groups choose or replace servers. The manual also explains request logs, configuration updates, scripts and HTTPS inspection using a trusted certificate. Using those features requires the application, a compatible Apple device and separately arranged network services, because the repository supplies no servers. The maintainer states that the project is unofficial and that behavior depends on the application version, so the manual does not establish tested compatibility or permission to reuse its text and configuration.
License
No license No license file found. Without a license, reuse of the code is not permitted by default.
Repository Website
8622026-10-29
haydenschmitty/floorplan
What it does
Floorplan is a Claude Code and Codex plugin for turning code structure into an explorable map. It targets builders trying to understand AI-assisted applications and developers joining unfamiliar projects. Repository files, imports, server routes and web requests become a local HTML map with relationship graphs and explanations. Node.js scripts discover files and links, while the AI reads code and describes components. Users can follow screens through servers and data, inspect recent changes and explore question-specific flows. It requires Node.js 18 or newer and authenticated Claude Code or Codex, followed by an explicit request to create the project's map. Updates rescan changes and retain unchanged descriptions. Request-to-endpoint matching is pattern-based rather than a compiler or runtime trace, so connections can be missed. Maps open locally, but description writing and questions can send code through the existing AI account. Its secret and exposed-endpoint checks are preliminary aids, not a security audit.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository Website
8612026-10-29
interplains/Zero-Breach-Toolkit
What it does
Zero Breach Toolkit is a defensive local checker that compares device evidence with published spyware indicators. It targets people examining their computers or phone backups for known traces. Inputs include browser records, selected operating-system artifacts, file fingerprints, AndroidQF collections and local iPhone evidence. Readable and JSON reports show matches, evidence strength and coverage gaps. Ordinary scans preserve source evidence, use temporary database copies and block Python socket calls. That guard is not operating-system network isolation, and indicator updates are a separate online action. Basic execution uses Python 3.8 or newer, while encrypted iPhone backups require an additional cryptography package and its supported environment. Optional Android acquisition requires owner consent and can change device state, unlike read-only scanning. No matches do not prove safety, and matches do not automatically establish infection. Full reports and unencrypted evidence ZIPs can contain sensitive records and need restricted sharing.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository
8602026-10-29
github/lever
What it does
Lever is a Copilot plugin and skill collection for measurable improvement loops in coding agents. Developers describe problems such as slow tests and receive a planning card. A Loop Card records metrics, measurement instruments, unchanged-run noise, baseline values, acceptance criteria and stop rules. Other skills cover experiments against controls, runnable briefs, divided refactoring work, PR review and progress reporting. It is installed in Copilot, with read-only card creation available in a new session. Personal and repository settings define autonomy and model-spend levels, subject to repository protections. Moving repeatable work into automations requires GitHub MCP tools; automations are created paused and tested before activation. Installation does not itself start universal automatic merging or scheduled execution. The project identifies itself as early, and useful improvements and safe delegation still depend on real instruments and permission settings.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8592026-10-29
amosblomqvist/learnharness
What it does
LearnHarness is an AI learning application that connects a learner's present understanding to a target through lessons and practice. Topic goals, answers, confidence and reasoning inform prerequisite checks, plans, explanations and questions. It accumulates propositions with their premises and conclusions in a knowledge graph rather than storing only right-or-wrong outcomes. Earlier sessions can inform later teaching and practice, and personal notes can be imported. Official architecture documentation separates extraction from teaching prose, identity review against existing knowledge and deterministic storage merging. The graph-tool source also exposes different information to diagnostic and extraction-review roles. Desktop releases for macOS, Windows and Linux start with provider API keys and model selection. Dictation and web search need separate service keys, and configured model providers receive task context. Chats, graphs and settings are local files, not proof of entirely offline learning. Neither reasoning accuracy nor educational effectiveness is established by the project description alone, so model output still needs checking.
License
AGPL-3.0 Network copyleft. Offering a modified version as a service also requires releasing the source. Review before any product use.
Repository
8582026-10-29
huaizhuanghub/dsh-persona-dafeiyu
What it does
This is a persona library plugin for DeepSeek Harness, not an independent AI model. Users choose bundled personas or import their own through Markdown, text files or pasted text. The selected body becomes a separate system prompt section, with changes taking effect on the next request. It supports profile-wide, workspace and session scopes, preferring more specific selections. Settings offer renaming, enabling, deletion, export and body inspection, with editing and deletion restricted for bundled entries. Bodies are fetched individually on demand rather than included in normal status metadata. It requires Node.js 20 or newer and DSH, with the npm package added to the intended profile. Global and workspace choices persist, but session choices stay in memory and disappear after restart. Switching personas needs no restart, while plugin code updates require restarting or refreshing the client. It does not manage execution permissions, and UI slot compatibility depends on the host version.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8572026-10-29
AienTech/dossier
What it does
Dossier is a self-hosted resume-writing and application-tracking app for an individual job search. Career profiles and job postings produce requirement comparisons, tailored resume and cover-letter drafts, and PDFs. Users can import resumes, refine documents in chat and track application stages and previous versions. The project describes a design that uses profile facts and shows missing experience as gaps rather than inventing it. Its Docker Compose example combines the app with PostgreSQL and defaults to local access. AI processing can use local providers such as Ollama or LM Studio, or separately configured hosted APIs. Job search, Gmail, Calendar and Claude connections each require setup and authentication rather than automatic activation. The optional job agent finds postings and prepares drafts but does not apply automatically, and email sending requires confirmation. It assumes one user and one profile with an English desktop-oriented interface, not a multi-user recruiting system. Hosted models receive relevant document content, so self-hosted storage does not eliminate external processing.
License
AGPL-3.0 Network copyleft. Offering a modified version as a service also requires releasing the source. Review before any product use.
Repository Website
8562026-10-29
milind-soni/OpenMausBot
What it does
OpenMausBot is a local-first agent application that manages several AI runtimes through one chat interface. Developers and individual users assign models and working folders to bots and request work in natural language. Conversations display file work, runtime events, approval questions and results. Official documentation describes a local harness server that owns agent processes and normalizes their events. Desktop releases embed that server, but agent CLIs and their authentication must be set up separately. Source execution requires Node.js 24 or newer and pnpm, with engines such as Claude, Codex or Grok connected. Boat cloud computers and Composio app connections require their own accounts or configuration and are not all activated by installation. An external MCP client can inspect teams and dispatch work but cannot grant approvals, delete data, change credentials or manage computer lifecycles. Local storage does not mean processing without external transmission, and Ubuntu Wayland host control is disabled. The main code uses Apache-2.0, while the enterprise directory has separate terms.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository Website
8552026-10-28
unclebob/ub-mode
What it does
ub-mode is an instruction skill for developers using coding agents, not a standalone application or model API. It addresses code changes that appear complete without demonstrating behavior, controlling complexity or examining weak tests. Inputs are a change request and repository context, while intended outputs are verified small changes and an explanation of the work. Its workflow starts with a failing test and a minimal working change, then calls for root-cause repair and behavior-preserving restructuring. The skill links separate tools for complexity and coverage, structural duplication, and mutation testing, which deliberately changes code to see whether tests detect the change. The skill's CRAP limit combines complexity and test coverage, and its duplication rule is a project policy rather than proof that similar code is wrong. Use requires a compatible agent skill loader plus separately available tools and the target project's test environment; the companion mutator declares Python 3.11 or later. Loading the text alone does not install or execute those tools, and neither metric thresholds nor agent instructions establish correctness or safe permission handling.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8542026-10-28
CopilotKit/OpenDuet
What it does
OpenDuet is a self-hosted proactive assistant template for people who need help prioritizing mail, preparing their day and following up on work. Connected mail, calendar information and user requests become morning briefs, task cards, drafted materials and actions submitted for approval. It builds on CopilotKit's OpenBot platform, not the unrelated nightly-labs/openbot desktop project. The server combines triage and routines with the CopilotKit runtime, and can connect agents through AG-UI, a protocol for exchanging agent interaction events. Each agent can use a container with its own browser, files and shell, while granted MCP connectors provide access to external applications. The documented gateway checks policy and records actions, and the policy source resolves conflicting rules toward stricter handling and requests approval when automatic review fails. The same source requires human handling for recognized credential changes, security changes and payments, so an approval screen does not mean unrestricted automatic execution of sensitive work. The documented starting environment needs Docker, Bun 1.3 or newer, Node.js, a model key, PostgreSQL infrastructure and a required CopilotKit Intelligence project for conversation threads and memory. Gmail and Calendar additionally require administrator OAuth configuration and each person's account connection, while agent instructions, skills and model settings live in a tenant package. This is an early project without a published OpenDuet hosted deployment guide or image, and self-hosting does not remove its required Intelligence service dependency or independently establish security.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8532026-10-28
google/elephant-goldfish-model
What it does
Elephant-Goldfish Model is a coding-agent plugin for teams that need to clarify design decisions before implementation, not a new language model. It takes requirements, design drafts and referenced source material and produces concept briefs, specifications and review reports. An Elephant keeps the project context, while Goldfish reviewers examine supplied documents without the earlier conversation. Review instructions separate broad architectural concerns, comprehension, design criticism and implementation readiness, with people deciding which findings to address. A separate skill compares design documents with referenced code and reports differences rather than silently repairing them. The plugin uses its host's models and permissions and requires the complete skills, reviewer personas and document standards, not just a copied skill file. Supported sessions must actually expose fresh subagent contexts for review, and the review stops when that capability is absent. The published verification record distinguishes tested local workflows from unresolved cloud trace checks, and the project explicitly does not replace tests or human judgment or constitute an officially supported Google product.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository Website
8522026-10-28
nightly-labs/openbot
What it does
This OpenBot is a local-first desktop workspace for people coordinating persistent AI teammates, and is unrelated to the CopilotKit OpenBot foundation used by OpenDuet. User requests and attached files enter separate agent conversations and workspaces, producing replies, files and delegated work. It connects local Codex and Claude runtimes and additional providers through ACP, a common protocol for connecting agent clients and runtimes. Electron owns the local processes and browser host, a SolidJS interface presents work, and SQLite stores the command log and derived views. Persistent message queues, agent messaging, managed file transfers and shared channels support continued work beyond a single chat turn. Queue control code distinguishes ordinary waiting messages from channel-owned tasks, and checks the active session before steering a message into running work. Packaged releases support macOS, Windows and Linux and use a supported provider runtime and its authentication, while source development additionally requires Bun 1.4.2 and Node.js 24. Local-first does not mean offline, because provider requests, visited pages and optional services can use external networks. The README and security policy describe a development preview with unrestricted agent execution and no routine per-action confirmation after initial consent, so separate workspace folders do not establish host isolation. The current license is PolyForm Noncommercial 1.0.0, and the README states that commercial use requires a separate license, making company adoption a permission review rather than an assumed open-source entitlement.
License
Custom license Custom or non-standard license. Read the license file before any reuse.
Repository Website
8512026-10-28
LiberNT/LiberNT
What it does
LiberNT is a ReactOS-derived operating system targeting Windows 11 compatibility for operating system and driver developers. NT means the kernel, driver model and system services behind Windows. Users supply applications and device inputs, while the system manages processes, files and display output. The source separates the kernel, hardware abstraction, drivers and user libraries, and its CreateProcessW implementation delegates process creation to an internal function. The README lists amd64, arm64 and riscv64 images, but compatibility is a development target, not evidence that arbitrary Windows software works. Testing starts with an ISO or preinstalled disk in a virtual machine or on spare hardware. Source builds use RosBE, CMake and Ninja on Linux, macOS or Windows through WSL2, with menuconfig selecting architecture and features. Connections are operating system interfaces rather than an AI service API, and the distributed Debug prereleases are explicitly unsuitable for a computer relied on for daily work.
License
GPL-3.0 Copyleft. Distributing software that includes this code requires releasing the source under GPL.
Repository Website
8502026-10-28
Orkas-AI/Orkas
What it does
Orkas is a desktop application for people who want AI agents to carry ongoing project work from a goal to reviewable results. It takes requests and project files and produces work such as reports, office documents and code. A Commander can delegate to named specialists, hand over a reply, or request an isolated anonymous worker. The inspected source defines separate concurrency limits and session locks rather than allowing unrestricted simultaneous execution. Project context persists across conversations, while the Library exposes file listing, passage search and source reading for global or project documents. The source setup requires Node 22.12 or later and a supported desktop system, with Python and a C/C++ toolchain needed for some native dependency builds. Users connect their model provider credentials or a local compatible endpoint, and external CLI agents require their own installed environments. Local storage does not mean cloud model requests stay on the computer, and collaboration, cloud synchronization and web publishing are not included in the open-source build.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository Website
8492026-10-28
ryoiki-tokuiten/Deepthink
What it does
Deepthink is a multi-agent exploration application for researchers and developers who want to compare different approaches to a difficult problem rather than accept one generated answer. It takes a problem and optional attached files, then produces candidate solutions, critiques and a selected final result. A strategy generator exchanges proposals with a proximity reviewer to reduce overlap before branches start. The implementation runs branches in parallel, waits at iteration boundaries, and repeats solution correction followed by separate critique. Optional periodic module work isolates smaller uncertainties, gives them to independent tester agents and routes their findings to relevant branches. Global updates can retain, evolve, roll back or replace branches, while compact memories preserve lessons and a final judge compares surviving results. A separate Adaptive Deepthink mode lets an orchestrator choose the next work pass without the fixed schedule or separate final judge. The repository supplies a Node.js based React and Vite application, provider configuration for API credentials or compatible endpoints, and an optional execution backend with sandbox settings that must be prepared separately. Its Vite configuration substitutes provider environment values into client definitions, so secret handling and exposure require review before shared hosting. More branches and iterations mean additional model calls, and neither a model judge nor the documented sandbox establishes independently measured accuracy or production security.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8482026-10-27
andyhuo520/digital-human-koubo
What it does
This repository combines agent instructions and scripts for making a talking-presenter video from an authorized photograph and a script or content request. It targets creators producing explainers, tutorials and product introductions, with an edited MP4 as the final output. Presenter references, voice descriptions and prompts are stored in persona.json, while build_tasks.py turns a script into generation-task JSON with estimated duration, resolution and credits. The documented workflow asks for approval of the quote and the first generated segment before generating the rest through Renoise MCP. That script prepares task data and does not itself enforce approval or submit generation requests. Subsequent steps compare whisper transcripts with the script, inspect sampled frames, clean up voices and edit captions and visual elements in Remotion. A complete run needs a skill-capable agent, a Renoise account and remote MCP access, plus local Python, Node, ffmpeg and whisper tooling. The text-behind-presenter effect additionally depends on macOS Apple Vision, so the same effect is not available through that path on every operating system. Account-specific reference materials must be replaced, generated speech and imagery need checking, and faces and voices require permission and AI disclosure. External generation costs and third-party rendering terms must be checked separately; published demonstrations and author-reported accuracy do not establish identical results in another environment.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8472026-10-27
t3dotgg/tstack
What it does
tstack is a collection of skills for coding agents that organize the work between a requested change and a reviewed GitHub pull request. It targets Claude Code and Codex users who want repeatable review, handover and PR monitoring procedures rather than a separate hosted service. Inputs are a change request or an existing branch or PR, and outputs include the changed code, verification evidence, review results and a PR handover or authorized merge. full-send links implementation, proof, sol-loop review, PR creation and CI monitoring, whereas full-stop ends with an open, unmerged PR for a person to handle. The documented merge gate checks the latest commit's evidence, required checks, review findings, branch freshness and mergeability. babysit-pr distinguishes T3 Code's watcher from the alternative pr-wait tool and requires bot findings to be checked against source before changing code. The README requires authenticated GitHub CLI access, python3 for pr-wait and access to other providers' models for council or sol-loop. These are editable workflow instructions, and model agreement or a watcher notification is not proof of correctness or permission to override an organization's approval policy.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8462026-10-27
enesozturk/dimensional-drawing
What it does
This repository provides agent skills for explaining a UI component's spacing, corner radii, layers and placement as a technical drawing rather than a screenshot alone. Its intended users include developers and designers handing over a component specification. Inputs can be source files, Storybook stories, URLs or screenshots, while the measurement procedure calls for reading the actual source and inspecting rendered boxes and computed styles with Playwright. The result is a self-contained HTML document with SVG plan views, exploded layers, details and variable tables. Dimensions use named variables and formulas, real copy is omitted, and unsupported values remain TBD instead of being invented. Drawing requires an agent that can read the skill and access the relevant source and browser environment. An optional motion skill targets Claude Motion, while its separate rendering path uses Node 18 or later, Playwright Chromium and ffmpeg to capture frames and encode an MP4. These are drawing instructions and rendering helpers, not evidence that a screenshot yields exact measurements or that the resulting component works correctly.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8452026-10-27
indexable-inc/blaze
What it does
blaze is a development build tool for Rust programmers who repeatedly wait for a program to rebuild after small edits. It accepts a Cargo project, source changes, and familiar build, run, or test requests, and produces development executables or test results. The initial build uses Cargo and rustc while a persistent worker loads the project and retains its analysis state in memory. Subsequent edits use rust-analyzer and salsa analysis, compile affected functions through a machine code backend such as Cranelift, and patch code into an existing executable rather than always relinking the entire program. Unsupported cases or changes outside the modeled inputs can return to Cargo, and the inspected state source records build guards and reasons for that decision. The documented accelerated path targets Apple Silicon macOS and Linux, while Windows builds use Cargo and Intel Macs are not supported. It uses the nightly toolchain pinned by blaze rather than honoring every project's toolchain pin, and retains extra memory and disk state. Live updates are optional and require a participating application, while changes to the shape of data already held by that application can require a restart. The inspected usage document warns that version 0.1.0 lacks borrow checking and main has incomplete coverage, so standard cargo check and CI remain necessary and published speed figures are not proof of equivalent behavior or gains in our environment.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository Website
8442026-10-27
every-app/every-app
What it does
EveryApp addresses the hosting, sign-in and permission work that remains after a coding agent creates an internal web application. It targets teams and developers who want to operate those applications in their own Cloudflare account. Inputs include an application definition, resource declarations and browser or AI tool requests, while outputs include a deployed application and authorized application results. A public Gateway checks access and forwards requests to private application Workers through internal connections, with signed identity verified at the application entry. The starter application scopes its database queries to the authenticated user's identifier, showing that row visibility remains the application's responsibility. Its tools declare separate read, write and destructive risk levels and are exposed to AI clients through the Gateway's MCP interface, a shared protocol for calling tools. Application declarations also specify allowed capabilities, while the Gateway keeps credentials and mediates external API access. The documented setup requires a Cloudflare account with Workers Paid and a Cloudflare managed domain, and the supplied stack uses Workers, D1, R2 and KV as applicable. The project explicitly remains in early development, so its access architecture is not proof of complete security, universal SaaS compatibility or production readiness.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository Website
8432026-10-27
StarTrail-org/LEANN
What it does
LEANN is a vector index and retrieval toolkit for users who want to search local documents and records without retaining a full stored embedding for every text chunk. It supports retrieval augmented generation, which means finding relevant source passages before asking a language model to answer. Inputs include documents such as PDF, text, and Markdown files, or records handled by the supplied source specific examples; outputs include matching passages and optional model generated answers. The document example reads files recursively with LlamaIndex and splits their contents into chunks before indexing. Its low storage approach uses a pruned graph and selectively recomputes embeddings, numerical representations of meaning, when the search needs them, with HNSW and DiskANN backend options. It requires a Python environment, an embedding model or endpoint, and a generation model for answers; native builds and optional scanned PDF extraction add platform and OCR dependencies. Models can run locally or through cloud APIs, so privacy depends on both the embedding and answer generation settings rather than the location of the index alone. Recomputing embeddings trades storage against computation and query latency, while extraction failures and model choice can affect what is actually found. The README's storage savings are author reported results under particular comparisons, not a guarantee for every collection or evidence of performance on our data.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository Website
8422026-10-27
current/deadcell-framework
What it does
This repository supplies the Dear ImGui foundation used by a separate deadcell menu, rather than the menu application itself. It targets C++ developers who need native windows and interactive controls. Application widget calls and operating system input become draw data, which the DirectX 11 backend submits to the GPU. The application repeats frame initialization, widget calls and rendering every frame instead of maintaining a retained widget tree. The build configuration combines the ImGui core and Win32 and DirectX 11 backends into a static library. The documented starting environment requires CMake 3.16 or later, a C++17 compiler, and an application providing a window and graphics device. The Windows build links Direct3D related libraries, while the backend source also uses the shader compiler. The public header identifies a fixed 1.89.5 WIP snapshot, so this is not evidence of current upstream feature coverage. The README says other bundled backends are not tested in this snapshot, and the downstream menu implementation is outside this repository.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository Website
8412026-10-27
gitroomhq/postiz-app
What it does
Postiz is a social media management application for teams that need to coordinate publishing across multiple channels. It addresses the work of preparing posts, arranging publication times, and reviewing channel analytics in one place. Users supply post content, media, connected accounts, and a publishing schedule, and receive scheduled or published posts and available analytics. Team members can collaborate and comment, while the public API and documented MCP and CLI interfaces provide entry points for external automation. Its documented stack combines a NextJS interface, NestJS services, Prisma with PostgreSQL, and Temporal, and the development configuration also defines Redis and supporting services. Users can choose the managed cloud service or operate the open source edition on their own infrastructure, but the inspected development compose file explicitly is not a production configuration. Self hosting requires storage and environment configuration, developer applications and approval from social platforms, and external model credentials for optional AI features. Self hosting therefore does not remove platform API permissions, third party processing, or the operator's responsibility for updates, backups, and security.
License
AGPL-3.0 Network copyleft. Offering a modified version as a service also requires releasing the source. Review before any product use.
Repository Website
8402026-10-26
designfuchen/word-atlas
What it does
Word Atlas is a web app for learning IELTS vocabulary through pictures and sound, with a reusable structured dataset. Learners answer recognition and spelling questions using word cards and examples, producing progress records and future review dates. Definitions, examples, British and American audio, and illustrations are connected through FSRS scheduling. Typing and dictation practice keep separate statistics without changing memory state. No account is required, progress stays on the device, and files support backup and restoration. Previously viewed material can be cached for offline review, but not every image or audio file is downloaded initially. Self-hosting requires Node.js 20 or later to build a static site, while JSON and CSV can be used separately. AI-generated material has automated checks but not complete human proofreading, and it is neither an official IELTS list nor a score guarantee. Code and learning content have different licensing terms, including attribution requirements for reused content.
License
Custom license Custom or non-standard license. Read the license file before any reuse.
Repository Website
8392026-10-26
rddmwmddr/android-ad-cleaner
What it does
This tool adjusts advertising and marketing-notification settings on selected Android phones from Chinese manufacturers. It identifies a connected device and its OS, then presents proposed changes and outcomes using settings screens and notification-channel information. It changes exposed vendor-app switches, while suspending advertising components requires the device owner's consent. This is distinct from rooting, unlocking a bootloader, or deleting system apps. ADB connectivity and USB debugging are required, with desktop, AI-assisted, and manual-guidance options. Desktop packages do not require a separate Python installation or AI tool, but device connection and selecting changes remain necessary. Backups and partial restoration are provided, while switches inside apps must be restored manually following the guidance. Paths and validation coverage vary by model and OS version, so universal Android support or complete ad removal is not guaranteed. The README distinguishes packaged fixes from source-only changes, making release contents important to check.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8382026-10-26
morganlinton/BeautifulRust
What it does
BeautifulRust provides three skills and lint configuration for coding agents writing, reviewing, or porting Rust. It is an agent-guidance plugin, not a standalone compiler or Rust model. Inputs include implementation requests, existing files or diffs, and source code with tests from another language. Outputs include Rust implementations, prioritized review reports with rule references and locations, or porting records mapping tests and deliberate differences. Review guidance separates error handling, public API design, ownership, and safety issues and supplies concrete fix examples. Porting guidance moves existing tests first to check behavioral differences. It must be connected as a Claude Code plugin or installed in a supported agent's skill location; installing skills does not enable project CI checks. Separate lint configuration must be incorporated into Cargo and run with a Rust toolchain to enforce machine-checkable rules. Small project-run evaluations do not guarantee correctness or defect removal across projects, so generated output and real tests still need review.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8372026-10-26
ludovic111/horizon
What it does
Horizon is an unofficial game project reimplementing Minecraft Java Edition behavior in Rust. It prepares legally owned game files and worlds as inputs and produces gameplay and saved state through its native client. A launcher handles installation, updates, and resource preparation, but the project does not supply original game code or assets. Use requires the specified Minecraft version, a JDK for initial preparation, and appropriate graphics support for the OS. Its own mod system does not run Java mod files unchanged, and servers requiring Microsoft account authentication remain limited. For agents, horizon-ctl reads game state as JSON and returns results after movement or interaction, with a separately connected MCP mode. Control requires a running game and local communication, so installation does not automatically connect other AI products. Project-provided differential tests cover compared behavior, not proof of complete equivalence in every situation. This was document research, not independent execution, compatibility, or hardware-performance validation.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository Website
8362026-10-26
JetBrains/thinkrail
What it does
ThinkRail is a development tool combining the pi coding agent with an editor and workspace management. Given a Git repository and a task, developers can inspect agent conversations, file changes, diffs, and execution records within one workspace. Separate git worktree branches and directories isolate concurrent tasks from the same checkout. A graph of SPEC.md files beside the code and reusable skills retain project intent and workflow context. Its architecture separates the agent host from the web interface and carries state over HTTP and WebSocket. Desktop packages and a browser CLI are available, but Git and model-provider authentication are required; installation does not grant model access. MCP servers need separate connections, with approval behavior based on read-only declarations and trust settings. Source development requires the specified Bun and Node.js environment, and planned full extension support must not be confused with shipped capabilities. Basic usage events and error reporting require privacy review before sensitive projects are used.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository Website
8352026-10-26
AgriciDaniel/claude-seo
What it does
Claude SEO is a Claude Code plugin and skill collection for inspecting website search visibility. Given a site or page URL, it analyzes technical settings, content, structured data, and AI-search readiness. Full audits combine specialist subagent results into prioritized improvement plans and report files with verification criteria. A headless browser handles JavaScript-rendered pages, and individual pages or topics can be audited separately. Beyond plugin installation, a supported Python environment and browser preparation are required, with diagnostic checks available. Basic audits are distinct from Google field-data and external search-data connections, which need appropriate accounts and API permissions. Multi-agent audits can increase model usage, and storing reports locally does not mean analyzed material avoids external model processing. Without field APIs, some performance and indexing judgments are estimates, while content revealed only after clicks or scrolling can produce noisy results. Findings do not guarantee search rankings or completed indexing and require follow-up with real pages and search tools.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8342026-10-26
apple/security-pcc
What it does
Apple publishes selected Private Cloud Compute components so researchers can examine its security and privacy properties. This is not a ready-to-install chatbot or general inference-server product. The repository contains security-policy components and research tools and points readers to an accompanying official security guide. The inspected attestation-verification CLI accepts a report or attestation-bundle file and dispatches it to a format-specific validator. Its output paths include verification summaries and details or validated key and expiry information. Certificate and transparency-proof settings are present, but inspecting them neither reproduces all PCC nor independently proves its security. Research-tool environments and software terms must be checked before use; cloning does not activate a cloud service. A custom license restricts use to internal PCC verification on one Apple computer for a limited period and prohibits redistribution, unlike ordinary reusable open-source licensing. The guide body was inaccessible in this retrieval, so this description is limited to inspected repository documents and source.
License
Custom license Custom or non-standard license. Read the license file before any reuse.
Repository Website
8332026-10-26
coollabsio/coolify
What it does
Coolify is a self-hosted platform for deploying and managing applications and databases on your own servers. Operators connect Git repositories, container images, or Docker Compose configurations to manage builds, deployments, and runtime logs. Workflows include deployment on Git changes, preview environments, and rollback to retained images. It also manages domains, HTTPS, container networking, health checks, scheduled tasks, and backup settings. The inspected production Compose configuration separates persistent storage for the platform, database, and cache. Getting started requires a supported Linux server, SSH access, firewall preparation, and storage. After installation, administrators still connect accounts, repositories, environment variables, domains, and permissions; existing apps do not all migrate automatically. Self-hosting leaves platform and workload security, capacity, and backup recovery with the operator. Listed backup or rollback features do not guarantee complete restoration of every application's data, so recovery procedures require separate testing.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository Website
8322026-10-25
aardappel/goose
What it does
Goose is a systems programming language and compiler for developers exploring compact data layouts and memory safety. It takes Goose source and emits C, which can become a native executable or run through the bundled TinyCC backend. Dynamic values live inline on compiler managed data stacks rather than a heap allocator. The compiler infers reference lifetimes, and scope exit releases storage. Examples include parsing order records, grouping customer totals and serializing relative references. Release builds support Windows, Linux and macOS, while native executable builds need an appropriate C compiler and source builds also require CMake and a C++20 compiler. The language is new, has whole program compilation and restricts escaping closures, recursive use of growable data and indexing variable size arrays. Its reported speed advantages come from repository benchmarks, whose summary also documents losses and cautions against treating aggregate speed and memory results as one implementation.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository
8312026-10-25
microsoft/skill-recorder
What it does
Skill Recorder is a desktop app that records work performed on screen and turns it into procedures an AI can reuse. It targets users and automation specialists who find it difficult to manually document repetitive work. Inputs include screen captures, application switches, window titles, clipboard previews, optional spoken narration and records from the app's own terminal. After recording, it groups events into step material, supplements ambiguous parts with relevant screen information and uses GitHub Copilot to reconstruct intent and sequence. Users can review and edit that analysis before generating an agent procedure file called SKILL.md or an automation procedure for scheduled or triggered execution. The resulting procedures are designed to prefer the target agent's command line tools and native capabilities over replaying clicks. Getting started requires installing the source release app and a GitHub account with Copilot access, with macOS the primary target and Windows 11 also supported. Recording and transcription are local, but choosing analysis sends events, extracted screen images and relevant text to GitHub's cloud. Preview and sensitive information protection do not guarantee safety or successful replay, so secrets must stay out of recordings and generated procedures need review.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8302026-10-25
elayadesign/ai-design-skills
What it does
AI Design Skills is a collection of written procedures for AI coding tools that can read Markdown rules. The current README lists one skill, landing-page-design, for creating a page around one offer, audience and primary action. Its inputs are the offer, audience problem, conversion goal, evidence assets and brand constraints. The procedure asks for an outline, headline copy, benefits, explanation steps, FAQ and search visibility guidance before code. It then specifies typography, spacing, colors, interaction states and motion rules for section by section construction. Starting requires a compatible AI tool and adding the skill file to that tool's project rules or skill location. It is a rule document, not a standalone page builder, analytics service or proof of higher conversion. Some visual prescriptions are rigid, and its examples of realistic numbers or names are not verified business evidence.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository
8292026-10-25
OpenMili/aimili-vpngate
What it does
AimiliVPN manages VPNGate nodes on a Linux virtual private server and provides a proxy gateway. A proxy gateway relays internet connections through another server. It retrieves node lists, checks availability and lets users switch connections. A web dashboard provides node and connection management. HTTP, HTTPS website access through CONNECT and SOCKS5 share one proxy port. The Python application uses the standard library, while tunnel operation requires OpenVPN and Linux networking components. The default proxy listener is limited to the local machine. The README recommends an SSH tunnel for access from another device rather than exposing that listener publicly. Volunteer-operated exit nodes have no guaranteed availability, speed, privacy or logging behavior.
License
Custom license Custom or non-standard license. Read the license file before any reuse.
Repository
8282026-10-25
metadist/synaplan-nextcloud
What it does
Synaplan Integration is a connector app that adds document summaries, translation and knowledge search chat to the Nextcloud file interface. It targets organizations already storing documents in Nextcloud and hosting providers operating those environments. It sends a selected file or question to a connected Synaplan server and displays a summary, translation or streamed answer. Uploading files into knowledge groups makes them available for RAG, which retrieves relevant documents to inform answers. Text extraction from binary documents such as PDF and Office files is handled by Synaplan's Tika functionality rather than by this app itself. Getting started requires a supported Nextcloud version from 33 through 35, PHP 8.2 or later, a running Synaplan instance and a valid API key. Operators can share one account, provision individual accounts or link existing Synaplan accounts, with individual mode keys stored encrypted on the server. Personal memories and image or video generation depend on the corresponding services and model configuration on the connected server. Changed files require a manual update through the dialog, while documentation and source distinguish automatic watched-folder processing as not yet implemented.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository
8272026-10-25
MapleTechLabs/maple
What it does
Maple is a platform for developers and operators investigating service failures and slow requests. It accepts traces, logs and metrics through OpenTelemetry, a standard for collecting application execution data. Users can inspect request waterfalls, search logs and explore metrics together. Hosted MCP access also lets compatible AI clients query telemetry. Its local mode combines an ingest server, query API and dashboard in a Bun binary that loads the embedded ClickHouse library chDB. Starting locally requires a supported macOS or Linux bundle and its companion library, while sending data requires an instrumented application or collector. Local mode is single tenant, and an unclean shutdown can require restoring a checkpoint with loss of later telemetry. The current license is FSL-1.1-ALv2, which restricts competing commercial services and provides a future Apache 2.0 grant rather than unrestricted Apache licensing today.
License
Custom license Custom or non-standard license. Read the license file before any reuse.
Repository Website
8262026-10-25
hugoj0s3/NaturalCron
What it does
NaturalCron is a C# library for defining recurring schedules in readable expressions. It targets developers adding periodic jobs or recurring reminders to .NET apps. Inputs are its defined English-like scheduling syntax or a builder that assembles conditions through code. It parses expressions into time-unit rules and calculates the next occurrence or multiple occurrences after a reference time. It supports weekdays, time ranges, exclusions, IANA time zone identifiers and developer-defined conditions such as holidays. The library only identifies execution times, so actual job execution, persistence and retries belong to a separate hosting service or scheduler. Getting started requires a compatible .NET project and the NaturalCron package, whose project targets .NET 8 and .NET Standard 2.0 and depends on TimeZoneConverter. An optional package generates traditional cron strings, but time zones, exclusions and custom keywords cannot be fully converted. It is not an AI that understands arbitrary Korean requests, and official documentation also forbids concurrent use of the same parsed expression instance across threads.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository Website
8252026-10-25
google-gemini/gemini-cli
What it does
Gemini CLI is a terminal based AI agent for developers working with code and operational tasks. It receives natural language prompts and project context, then can return explanations or use tools to inspect files, change code and invoke shell commands. MCP servers can supply additional tools, while GEMINI.md files provide persistent project instructions. Headless mode returns text, a JSON response with usage statistics, or streamed events for tool requests and results. The package requires Node.js 20 or later and a supported Google login, Gemini API or Vertex AI authentication setup. Sandboxing can restrict execution, but requires platform appropriate setup and does not eliminate every risk. The current README says unpaid tier and Google One users moved to Antigravity CLI on June 18, 2026. It also warns that Gemini CLI is not optimized for the latest Gemini models, specifically Gemini 4 Argon, and directs those users to Antigravity instead.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository Website
8242026-10-25
punitarani/fli
What it does
Fli is a library for querying Google Flights information from software. Its main users are developers building travel search and conversational assistants. Inputs include origin and destination airport codes, dates, passenger counts, cabin class and stop preferences. It returns prices, durations, airlines and departure and arrival details for individual legs. It provides a Python library, a command line interface and an MCP server, using the standard through which AI calls external tools. Although the introduction emphasizes direct API access, the current implementation fetches search pages and decodes the flight data embedded in them. Getting started requires Python 3.10 or later and the flights package, with separate optional dependencies for MCP. The current transport cannot perform multi-city searches or retrieve booking options, and some filters, including baggage, are unsupported. Date searches fetch one page per date and cover at most 93 dates per call, so an empty result does not establish that no flights exist.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository Website
8232026-10-24
xevrion/hatchzip
What it does
HatchZip is an archive extraction product for disks without enough space to keep both a large archive and all extracted files. It accepts ZIP, 7z, tar, tar.gz and split archives, creates an output directory and progressively reclaims processed archive storage. ZIP, 7z and plain tar are processed from the end, saving progress before truncating the source. tar.gz uses a different forward-reading approach that reclaims consumed ranges on supporting filesystems. GUI and CLI interfaces provide a non-mutating space check and interrupted-run resumption. This is a locally installed product whose core extraction does not require a model API or AI account. Because the original archive is progressively consumed, integrity checks and a separate retention policy must be reviewed before using the only source copy. A single large solid 7z block or an output disk different from the source limits space savings. Encryption-support statements differ between the README and design document, so that feature is not confirmed here; the actual release and recovery restrictions require checking.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository Website
8222026-10-24
makecindy/cindy
What it does
Cindy is an open-source client for an AI agent that works with local files and signed-in apps. Its desktop app uses Electron and its mobile app uses Expo and React Native; the backend is not included in this repository. Users provide tasks and select execution harnesses such as Claude Code or Codex together with models to obtain results. The README describes switching harnesses while retaining workspace, memory, skills and tools, alongside browser and device control and message or schedule inputs. Models require a separate connection through the official service, existing coding-plan authorization, personal API keys or local models. Skip Sign-In permits local agents but neither enables server-backed capabilities nor connects to a self-hosted server. Source development requires Node.js 22.x, pnpm 10.x and Git LFS, and default endpoints point to the official cloud. The plugin marketplace and team skill sharing are marked as in development rather than complete. Official builds disclose usage analytics, so local execution does not mean that no data ever leaves the device.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository Website
8212026-10-24
anthropics/sandbox-runtime
What it does
Anthropic Sandbox Runtime is a CLI and library that restricts file access and networking for arbitrary processes, including agents and local MCP servers. It accepts a command plus filesystem and network policies and executes the command within operating-system boundaries. macOS uses sandbox-exec, Linux uses bubblewrap, and HTTP and SOCKS5 proxies check permitted network destinations. The current README also describes alpha Windows support using a separate account, restricted tokens, file ACLs and WFP network filters. The package requires Node.js 22.12.0 or later, and restrictions apply only when commands are explicitly wrapped or the library is integrated. Linux requires bubblewrap, socat and ripgrep, while Windows requires a separate elevated setup step. Default reads are allowed, so automatic protection of secret files must not be assumed. Read-denied paths, write-allowed paths and allowed domains must be configured for the workload. This is an early research preview, and documented Windows gaps involving alternative tokens and service-mediated traffic prevent treating it as an absolute security boundary.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository
8202026-10-24
humanlayer/humanlayer
What it does
HumanLayer is an execution product for teams managing coding agents across their own machines. Developers submit tasks and messages, and a host daemon runs Claude Code, Codex, OpenCode or Pi sessions while connecting tasks, plans and reviews to the app. The repository contains a web interface, macOS desktop apps, a CLI, a daemon, an API and shared packages. It separately includes an Electric authentication proxy for syncing Postgres data to clients and plugins for connecting existing sessions. Remote use requires host authentication or a short-lived launch token and a running daemon, not merely CLI installation. Launching the daemon does not create a persistent background service, so restart supervision and credential persistence need separate preparation. Source development requires Bun and Docker, with Rust additionally required for the Tauri app. Self-hosting guides are explicitly still in progress, so public source should not be mistaken for a turnkey production installation.
License
Apache-2.0 Permissive, with a patent grant. Commercial use is allowed; keep the notices and state your changes.
Repository Website
8192026-10-24
nodejs/node
What it does
Node.js is an open-source runtime for executing JavaScript outside a browser, not a finished business app or AI model. It provides the foundation for server and CLI developers to process scripts, files or network requests and produce responses, files and other outputs. Its V8 engine and asynchronous I/O APIs allow other work to proceed while file or network operations are pending. The official introduction demonstrates the input and output flow with a small HTTP server receiving requests and sending responses. Getting started requires an operating-system-compatible runtime and application code; external API authentication is connected separately only when the application needs it. Installing the runtime does not create business automation or a service. The project distinguishes release types and support periods, so deployment versions and package compatibility require checking. Asynchronous I/O does not automatically parallelize CPU computation, so long-running calculations need separate design to avoid blocking the event loop.
License
Custom license Custom or non-standard license. Read the license file before any reuse.
Repository Website
8182026-10-23
oosexgang/hackios-xnu-kernel
What it does
HackiOS is an operating-system porting research project that combines a modified open-source XNU kernel with a separately prepared iOS 16.5 userspace on Android hardware or QEMU. This repository contains kernel patches, an EFI loader, platform support, build helpers and some prebuilt components, not a complete independent mobile OS. Its documented inputs include a companion userspace repository, an image prepared from Apple's IPSW and, for physical hardware, the device's own firmware. The reported targets are the POCO X5 Pro 5G and an aarch64 QEMU virtual machine on Linux, with timed research sessions and diagnostic logs as part of the workflow. The README reports selected interface, Wi-Fi, audio and camera functions, but cellular service, Bluetooth, charging under this kernel and Apple ID or App Store content are not available. Inspection of miniamfi.c confirms that the replacement AMFI interface has no ordinary signature policy behind it, so this is not equivalent to a supported iPhone security environment. It is not a daily-use phone ROM, and the kernel's open-source licensing does not grant rights to Apple's userspace or device firmware.
License
Custom license Custom or non-standard license. Read the license file before any reuse.
Repository
8172026-10-23
Revivedx/SavageWindRecomp
What it does
Savage Wind Recomp is an experimental static recompilation project for running the Switch version of Breath of the Wild as a native PC application. Its audience is testers who already have their own legally obtained and extracted version 1.8.2 game files. The initial setup accepts ExeFS and merged RomFS folders, checks the version, builds a data image and compiles game code locally. The project combines a recompiler with replacements for console services and a Vulkan graphics runtime, rather than supplying the game itself. The Windows package requires Windows 10 or later, an AVX2 CPU, Vulkan 1.3 graphics and substantial free storage; Linux support remains less tested. The README reports testing only the opening area and Great Plateau, so its performance claims do not establish full-game compatibility. DLC and several graphics features remain planned, and neither the project's code license nor its installer grants rights to redistribute Nintendo content.
License
Custom license Custom or non-standard license. Read the license file before any reuse.
Repository
8162026-10-23
lemomo-ai/lemo-opuscar
What it does
Lemo-Opuscar is a style library, production toolkit and agent skill for making short films with code rather than a video-generation model. A creator supplies a topic, style and any owned materials, then a coding agent develops scenes, a timeline, narration, music and an exported film. The technical guide describes deterministic Canvas or WebGL pages captured frame by frame in a headless browser, with speech and sound mixed and muxed through ffmpeg. The plugin manifest is an entry point to those guides and tools, not a standalone renderer that starts producing films immediately after installation. The documented environment requires Node 20 or later, ffmpeg and Python 3.11 or later, with browser packages, speech models and assets added as needed; Windows use is directed to WSL. English narration can use local Kokoro, while the documented Chinese Edge TTS path needs network access. The author tunes the styles for Claude and warns that other models may not reproduce them reliably, so output still needs visual, sound and factual review. Code and documentation, demo films, third-party assets and several excluded voice tracks have different reuse conditions that must be checked separately.
License
MIT Permissive. Commercial use and changes are allowed if the copyright notice is kept.
Repository Website

Showing 50 of 962

Only repositories in the ranked Trendshift lists are included, and the lists are used only to find candidates. We do not copy their ranks. Descriptions, licenses and star counts come from each GitHub repository. The notes are our own reading. We have not tested these projects, and a place on a trending list does not prove quality. Trendshift RSS feed